Keep this open while facilitating
The workshop in five handoffs
For the simple visual view: Open the two-day Workshop Flow Map. It covers all 34 operating steps, the three sites, who submits, and the next button to press. This guide keeps the longer teaching scripts.
| Who | Next action |
| Facilitator | Present the deck; in Live Console → Run, select the activity and press Launch on phones. Selecting alone only previews. Choose Screen: Live activity; set a timer. |
| Students | Discuss together; COO submits one team answer. Everyone submits their own polls, surveys, networking completion and 90-day plan. Wait for server confirmation. |
| Live app | Save answers and show response counts. Open accepts answers, Lock stops them, End makes phones wait. A timer is a cue; use Lock at the deadline. |
| Captain | Review your region’s six teams, add judging evidence and Approve for facilitator. Approval alone does not add credits. Captains can award reviewed credits once within their region, or use the reasoned any-activity award form. Captains cannot launch global activities or score another region. |
| Facilitator | Reviewed activities: Workshop → Captain approvals → Apply once. Quizzes: Run → Reveal and score. Cloud or Not awards +100 per correct participant and aggregates their team score; recap and service-model quizzes award +10 per correct team. Mystery, Poker, Bill Shock and vote panels have their own score buttons. End and transition. |
Use the right site
Workshop Pages: decks, scripts and this guide. ChaiCart Live / Firebase: Google join, answers, reviews, credits, awards and certificate-portal access. ChaiCart Demo / Azure: ordering, failure and recovery demonstration; Demo orders do not earn workshop points.
Before / between days
Create one Live session; Setup generates all team codes, which captains share. Project the Live Screen and keep Console and Demo controls private. Approve and freeze every team’s architecture before Day 2, including non-winners. Resume the same session on Day 2 so credits and designs carry forward.
Scoring guardrails
Use only one scoring path per award. Do not award a quiz through captain reviews or award a Shark regional win through both votes and pitch reviews. Networking stays open through Workshop, not a Run launch. Permanent certificates are issued per verified attendee in InspiredLearning after final scoring and workshop closure. The Live certificate toggle announces the portal only. The old static quiz/leaderboard pages are optional reference, not the live scoring system.
Start here
How the workshop works
The story
Every team is the founding team of ChaiCart, a 10-minute chai and snack delivery startup. Each topic on the agenda becomes a decision the startup has to make:
- Day 1, Build: choose a service model, a deployment model, a cloud provider and an architecture.
- Day 2, Survive: launch day goes wrong. Teams investigate an outage, manage reliability, control costs, connect business systems and pitch at Demo Day.
Refer back to ChaiCart in every explanation: “So what would this mean for ChaiCart?” The story is what makes the concepts stick.
Structure at a glance
| Unit | Size | Named after |
| Student | 120 (2nd, 3rd and 4th year) | — |
| Team (startup) | 24 teams of 5 | Availability zones: Mumbai-1a … Delhi-1f |
| Region | 4 regions of 6 teams | West India (Mumbai), South India (Chennai), Central India (Pune), North India (Delhi) |
The team → region structure mirrors real cloud design (availability zones inside regions) and lets you run 24 teams without the day dragging: pitches, judging and marking happen inside regions in parallel, and only the region winners come to the main stage.
Mix the years
Include 2nd, 3rd and 4th year students in each team where possible. Share project and internship experience, and rotate who explains the team's reasoning. Pre-assign teams on a list and project it as students arrive.
The five roles
| Role | Owns | Used in |
| CEO | Final decisions, presenting | Shark Tank, Demo Day |
| CTO | Frozen app architecture | Architecture Lego; proving the design in Error Budget Poker and Bill Shock |
| SRE | Error budget tracker | Murder mystery evidence lead, Error Budget Poker, postmortem |
| CFO | App credit ledger | Bill Shock, checking the leaderboard |
| COO | Time, materials, forms | Submitting team answers and coordinating the Digital Delivery Factory |
Cloud Credits
Each team starts with 1,000 credits when its Live session is created. The official scores are stored in the ChaiCart Live credit ledger. Project its Screen → Leaderboard at transitions. Captains approve reviews and can apply them once within their region; facilitators can also apply them, and dedicated panels score quizzes and games. leaderboard.html is a legacy practice tool, not the workshop ledger.
Your rhythm for every block
- Hook (1–2 min): a question, a story or a surprise.
- Concept (5–10 min): the analogy slide first, then the technical slide.
- Activity (10–30 min): teams play; captains circulate in their regions.
- Debrief (5 min): ask the debrief questions before revealing answers.
- Score and transition: captains save reviews digitally; captains or the facilitator apply approved credits; dedicated scoring panels score automatic activities, then reads the transition line.
Three networking icebreakers
Day 1 opening: LinkedIn (slide 6, 7 min). Day 1 after lunch: GitHub (slide 26, 7 min). Day 2 opening: X (slide 4, 5 min). These fit within the existing session blocks; keep introductions and debriefs brief.
Students create or improve their profiles, connect with peers, and use the presenter’s QR codes. LinkedIn: request connections with peers and Vishruth. GitHub: follow peers and theharithsa. X: launch a learning circle in the opening, then continue sharing photos and takeaways throughout Day 2; posting is optional. Use #ChaiCartCloudWorkshop on workshop takeaways and photos so everyone can find them through the hashtag search. Existing accounts count.
Scoring: +100 Cloud Credits per team per activity, once only; maximum +300 over both days. Every teammate completes the activity, and the COO reports completion to the captain. Existing accounts, connections and follows count; students improve their profile and make a useful peer introduction. Do not collect passwords, OTPs, profile screenshots or lists of personal handles. If account setup is delayed, students draft their profile and finish during a break; award the activity once completion is reported.
Review the three networking activities in the captain dashboard, then apply each team review once from Workshop. QR codes are local SVG assets; opening profiles still requires internet.
Using the slide decks
- → / Space: next (reveals items one at a time first). ←: back. F: full screen. Home / End: first / last slide.
- Activity slides have a timer: click Start. It turns clay-coloured when time is up.
- Jump to a slide by adding
#N to the address, e.g. day2-slides.html#13.
Logistics
Room, regions and captains
Room layout
- 24 tables of 5, grouped into four quadrants, one per region, with a clear aisle between quadrants (leave space for volunteer demonstrations).
- A stage area at the front with space for 7 people in a line (Human Kitchen, Follow the Order).
- Space for small group discussion during the digital gallery rounds.
- Project the team map and tell each table its team name and join code.
- Projector plus two microphones: one for you, one handheld for volunteers and pitches. In a room of 120 a mic is not optional.
People you need
| Role | Who | Does |
| Lead facilitator | You | Runs the slides, explains, debriefs, keeps time. |
| Region captains ×4 | Faculty members or senior volunteers | Run region rounds, review digital answers, and submit approved reviews. Share the captain briefing and digital review reference. |
| Scorekeeper | One volunteer with a laptop | Helps verify reviews and project the Live leaderboard; credit application requires an authorized facilitator account. |
| Tech helper (optional) | A volunteer or lab assistant | Keeps Azure, Dynatrace and the web app running; handles the second screen. |
Each captain account covers one region. If you have only two helpers, assign them one region each and use the facilitator console to review the other two regions and apply scores. Without any helpers, the region rounds still work: teams vote, and region results are read out by one elected CEO per region.
The region-round pattern
Used for Shark Tank, Architecture Lego, the postmortem and Demo Day:
- All 24 teams do the activity at their tables.
- Inside each region, teams present briefly to each other (60–90 seconds each) while the captain times.
- The region votes (each team one vote, not for itself) or the captain picks the best.
- The four region winners present on the main stage. You and the captains choose the overall winner.
This keeps all 24 teams active, takes about the same time as 4 presentations, and gives every region a moment on stage.
Before the workshop
Preparation checklist
One week before
- Get the student list and assign 24 teams of 5, mixing 2nd, 3rd and 4th year students. Assign 6 teams to each region.
- Confirm with the organiser: hall for 120 with movable tables, projector, 2 microphones, wall space, Wi-Fi.
- Confirm 4 region captains and a scorekeeper. Send them the captain briefing.
- Prepare reliable Wi-Fi, phone charging, two microphones, projector, four captain accounts and prizes for the closing awards.
- Run through both decks once, out loud, with a timer. Rehearse the Azure and Dynatrace demos.
Paperless student and captain setup
ChaiCart Live is the student workspace. This is separate from the Azure storefront used for cloud demos.
- Open Console, sign in as a facilitator, create or select the session and set its workshop date. Setup displays the 24 team join codes.
- Register four captain Google emails with one region each in Setup. Captains open Captain dashboard; they can review only their six teams.
- Students sign in with Google, enter the session and team code, and claim one of five unique roles. The COO submits shared answers; everyone completes their own surveys, networking and 90-day plan.
- Run selects and opens the live activity. Workshop provides Bingo clues, frozen designs, digital stations, factory rounds, review approval, X shortlists and certificate release.
- Keep Networking open across the day for optional X posts using #ChaiCartCloudWorkshop. Ask permission before photographing people. Judge learning, creativity and workshop spirit, not likes. Record the individual award before releasing certificates.
No printed cards, worksheets, envelopes, dice or planes are required. Human Kitchen and Follow the Order still use volunteers, with digital station cards and timestamped handoffs. Keep Wi-Fi, chargers, a projector and microphones ready. If connectivity fails, continue the discussion verbally, pause submissions and record corrections after service returns.
Operating and recovery instructions: ChaiCart Live README. Legacy printable packs remain optional reference material.
Deployed ChaiCart app — presenter setup
Storefront: chaicart-workshop-vh-20261003.azurewebsites.net
Facilitator console: Open facilitator controls
Runbook: App demo sequence · Deployment and recovery notes · Observability runbook
| Azure setting | Current environment |
| Subscription / region | D1/APAC / Central India |
| Resource group | rg-chaicart-workshop |
| App / plan | chaicart-workshop-vh-20261003 / plan-chaicart-workshop |
| Runtime / capacity | Node 24 LTS, Linux B1, one instance; HTTPS, Always On and /health check enabled |
| Order storage | /home/chaicart-data/state.json; keep one app instance |
- Use this deployed app for the PaaS and ordering demo; do not recreate the old “ChaiCart is live!” placeholder. Customers sign in with Google to order chai, coffee or snacks. Payments and delivery are simulated.
- Open storefront and facilitator console in separate tabs. Sign into the console with a verified Google account whose email exists in Firebase
admins/{email}. Customers need no admin entry. There is no public facilitator token.
- Before class, verify Google sign-in, one order, tracking, and facilitator controls. Select Healthy / Roll back, stop any surge, and prepare a QR code for the storefront. Health and public-route checks passed at deployment; authenticated flows and crowd capacity need rehearsal.
- Use a separate resource group,
rg-chaicart-day1, for the VM/container race. Open Cloud Shell beforehand and register Microsoft.ContainerInstance if needed. Keep the workshop app running between Day 1 and Day 2.
- The app has local metrics, structured logs and logical traces. Dynatrace RUM and OTel are deployed; all three OTLP signals passed an ingestion test. Rehearse a real signed-in flow and verify its tenant data before presenting. Application Insights is separate and is not established by the Dynatrace setup; use prepared evidence for unrehearsed tool tasks.
- The repository's ChaiCart demo — test and deploy workflow runs tests, but automatic Azure deployment is not enabled. Configure the deployment credentials and rehearse a green deployment before presenting CI/CD as live. See the deployment notes; a Git push alone currently does not update Azure.
- Backend Firebase credentials are stored in an encrypted App Service setting. Keep credentials, settings values and Google sign-in screens off the projector. Key Vault migration requires a subscription administrator. After the workshop, export evidence and remove the dedicated resources when no longer needed; stopping the app alone does not stop plan charges.
Dynatrace preparation
- Rehearse the deployed ChaiCart environment and confirm a real signed-in trace is visible. For unavailable hunt questions or a recent problem, prepare separate evidence and clearly label its source.
- Bookmark Services, Distributed Traces, Problems, Logs and a Notebook with the treasure-hunt queries already pasted in.
- Run every DQL query from the Day 2 slides and the worksheet; adjust field names if your environment differs.
- Choose one “story” trace in advance: a failed request with an obvious slow span.
Morning of each day
- Tables in four quadrants; team map, session code and join instructions projected.
- Check signed-in students, unique roles, captain assignments, Timeline, Sort and Bingo screens.
- Check frozen Day 1 designs, protected mystery evidence, treasure queries and Factory controls.
- Open the deck, ChaiCart Live console and session projector screen in separate tabs, plus the deployed ChaiCart storefront and facilitator console. Sign into the console, restore healthy mode and confirm an order. Have Azure and the verified Dynatrace environment signed in for Day 2.
- Use a fresh session for a new workshop; preserve the previous session and ledger.
| Day | Activity | How to score | Credits |
| 1 | LinkedIn networking | All teammates complete the profile and connection steps; existing connections count; once per team | +100 |
| 1 | GitHub developer circle | All teammates complete the activity; existing profiles and follows count; once per team | +100 |
| 2 | X learning circle | All teammates complete the activity; existing profiles and follows count; once per team | +100 |
| 1 | Human Timeline | First correct team in each region | +50 |
| 1 | Service Model Sort | Captain checks stored choices: 5 credits per correct card | up to +100 |
| 1 | Where Should They Live? | Each convincing defence (applause decides) | +30 |
| 1 | Cloud Shark Tank | Region winners · grand winner (extra) | +50 · +100 |
| 1 | Cloud Bingo | First three teams | +30 each |
| 1 | Architecture Lego | Region best · overall best (extra) | +50 · +100 |
| 1 | Human Kitchen | Each volunteer's team | +20 |
| 1 | Gallery Walk | Best digital case reflections (pick 4) | +50 each |
| 1 | Cloud or Not | Individual answer; Reveal and score adds personal credits to the team total | +100 per correct participant |
| 1 & 2 | Recap quizzes | COO submits; dedicated quiz controls score answers | +10 per correct |
| 2 | Guess the Downtime | Captain compares stored annual-minute guesses; closest per row | +20 |
| 2 | Murder mystery | Killer +200 · weapon +100 · accomplice +50 · fix +50 · Exhibit F +25 · first correct in the room +100 · wrong killer −50 | up to +525 |
| 2 | Treasure Hunt | Captain reviews all 15 stored answers and evidence (max 150) | up to +150 |
| 2 | Error Budget Poker | As shown in the app; leftover minutes × 3; budget below 0 → −100 | varies |
| 2 | Blameless postmortem | Region best · overall best (extra) | +40 · +75 |
| 2 | Digital Delivery Factory | Most successful round-2 deployments per region; ties use fewer failures, then lower lead time | +50 |
| 2 | Cloud Bill Shock | As shown in the app; selecting the correct preventing control refunds half a loss | varies |
How scores reach the leaderboard
- Teams read the credit ledger in My workshop (the CFO checks it).
- Captains save approved reviews with feedback for their six teams.
- Captains use Award reviewed credits to team; facilitators can apply the same review once in Workshop. Undo adds compensating ledger entries when correcting a batch.
- Use stored submissions and protected rubrics; never accept a self-reported total as the only evidence.
Awards at Demo Day
- Richest Startup: top of the overall leaderboard after the final quiz.
- Region Champions (×4): top team in each region.
- Sherlock Award: first correct murder mystery solution in the room.
- Best Architecture and Best Blameless Postmortem: the overall winners of those activities.
- Most Creative Pitch: from the Demo Day grand final.
Best Workshop Post or Photo: announce one individual winner at closing. Students submit their best public X post link to their captain before the final awards, using #ChaiCartCloudWorkshop. Captains shortlist entries; the facilitator judges learning value, creativity and workshop spirit, not likes or follower counts. Remind students to ask permission before sharing someone’s photo. This award is separate from the +100 team networking credits; public posting is optional.
Keep scoring visible and generous. With 24 teams, most students watch their region more than the overall table; the region view keeps every team in a race it can win.
Day 1 · Cloud Computing Fundamentals & Industry Practices
Build ChaiCart
09:30 · 25 min
Opening, agenda, LinkedIn and “Cloud or Not?”
Slides 1–8
- Slide 1: pause on the title, then reveal the line.
For the next two days you are not students. You are founders. You're going to build a company in the cloud, and tomorrow we find out if it survives.
- Slide 2: walk through today’s agenda and breaks; show when LinkedIn and GitHub icebreakers happen.
- Slide 3: use the three-site illustration: Workshop site for instructions, ChaiCart Live for student work and credits, and the Azure demo for simulated orders and incidents. Introduce Build → Survive → Pitch and the 1,000 starting credits.
- Slide 4: explain teams as availability zones and regions. Introduce the four captains by name and region; ask each region to cheer.
- Slide 5: start the 5-minute timer. Students sign in with Google, join using their team code and assign the five roles. Explain the illustrated handoff: COO submits team work, captain reviews and can apply the award to the ledger; the facilitator can apply it too. Cloud or Not uses individual answers at +100 each; recap quizzes use the COO’s answer at +10. Reveal and Score applies automatic points.
- Slide 6: LinkedIn icebreaker (7 min). Pair students across teams, scan the presenter’s QR, and make a useful introduction. Captains approve +100 reviews after all five members complete; the captain or facilitator applies each review once; existing profiles and connections count.
- Slide 7: everyone stands. Reveal items one by one: stand = cloud, sit = not cloud. Hand the mic to one student per item to justify.
- Slide 8: Use Reveal and score for each Cloud-or-Not question; award +100 per correct participant and add those credits to their teams. Check individual work under Students; discuss UPI and Excel as the two twists.
Debrief
- “What makes something cloud?” Look for: someone else runs it, network access, shared, pay for what you use.
120 students standing and sitting is loud and fun. Let it be, but keep it to about 8 minutes. The definition comes on the next slide.
“You just defined cloud better than most textbooks. Here's how NIST puts it.”
09:55 · 35 min
What is cloud, Human Timeline, why businesses move
Slides 9–12
- Slide 9: electricity analogy; reveal the five cards. For each, ask a random table: “What's the ChaiCart example?”
- Slide 10: Teams arrange the 12 digital event cards. No talking for the first two minutes; then discuss and submit. The app locks only a correct order. Captain checks server timestamps and approves the first correct team in each region; captain or facilitator applies +50 once; incorrect orders can be reviewed with zero-credit feedback.
- Slide 11: reveal the timeline year by year while teams check their line. Then the three eras.
We went from sharing one big computer, to everyone owning their own, and now back to sharing. Only this time the sharing is global and on demand.
- Slide 12: reveal the six drivers. Ask one team per region which driver matters most to ChaiCart.
1998/1999 and 2013/2014 cause most mistakes. That's useful: it starts the VMware vs Salesforce and Docker vs Kubernetes conversations.
10:30 · 30 min
Service models: IaaS, PaaS, SaaS
Slides 13–17
- Slide 14: Biryani-as-a-Service. Ask: “Who has ordered on Swiggy? Why not cook?” The answer is always control vs effort.
- Slide 15: map the analogy to the stack, one row at a time: “In PaaS, who patches the OS? The provider. That's why developers love it.”
- Slide 16: sorting race, 6-minute timer. Run → Service Model Sort → Launch on phones; teams classify the digital cards, and COO submits.
- Slide 17: reveal column by column. Captains compare the stored classification with the protected key; the review computes five points per correct card. Captain or facilitator applies the reviewed score once. Discuss the tricky ones: Azure Functions (serverless, PaaS family), Blob Storage (raw infrastructure), Dynatrace (SaaS).
Debrief
- “ChaiCart has 3 developers and no ops person. Which model for the app? For email? For the database?”
Break (15 min). “After the break: where should ChaiCart live?”
11:15 · 25 min
Deployment models: “Where Should They Live?”
Slides 18–21
- Slide 19: hotel, own house, house + Airbnb, many hotel chains.
- Slide 20: captains assign one digital scenario per table, so three teams share each scenario. 3-minute timer to decide.
- Go through the eight scenarios. For each, ask the three teams to raise hands, pick one at random for a 60-second defence (handheld mic), then let one of the other two challenge with a single question.
- Best debates: bank (regulation vs innovation), hospital (sensitive data + AI), factory (latency → edge).
- Slide 21: debrief the six factors. Mention RBI data localisation and the DPDP Act as real Indian constraints.
Don't present the suggested answers as “the truth”. Say: “Here's what most architects would suggest, but your argument counts more than the label.”
11:40 · 45 min
Major platforms and Cloud Shark Tank
Slides 22–25
- Slides 23–24: the landscape and the “Rosetta Stone”. Quiz the room by pointing at a row and asking for the Azure name.
- Slide 25: teams open their assigned provider card and the client brief above the Shark Tank pitch form. 8 minutes to prepare.
- Region round (10 min): in each quadrant, every CEO pitches for 1 minute while the captain times. Each team then votes for the best pitch, not its own. The captain announces the region winner.
- Grand final (15 min): the four region winners pitch for 2 minutes on stage. You ask each the “They will ask” question from their card. You and the captains pick the grand winner.
Be strict at 1:00 in the region round and 2:00 on stage. Cutting pitches off politely is part of the fun.
Lunch. “After lunch we wake up with Bingo, then open up the machines.”
13:15 · 40 min
GitHub icebreaker, Cloud Bingo, virtualization and containers
Slides 26–31
- Slide 26: GitHub networking (7 min). Improve profiles, follow peers and the presenter, and discuss one shared project idea. Captains approve +100 reviews after all five members complete; the captain or facilitator applies each review once; finish delayed signups during a break.
- Slide 27, Bingo (8 min): read clues from the Workshop clue caller in random order, ticking each one. The COO submits a Bingo claim; captains check its digital line against called terms. Confirm the first three valid room-wide claims before applying credits.
- Slide 29: house → apartment → tiffin box. “In an apartment, what's shared and what's yours? In a tiffin service?” (The kitchen = the kernel.)
- Slide 30: walk both stacks from bottom to top; reveal the Kubernetes line.
- Slide 31, live demo: start a VM in the portal (in
rg-chaicart-day1), then immediately run the container commands in Cloud Shell. The room bets on which serves a page first. Open the container's IP on screen.
- After the session:
az group delete -n rg-chaicart-day1 --yes.
If Wi-Fi is slow, show a screen recording. Never debug live for more than 60 seconds.
13:55 · 50 min
Cloud architecture and Architecture Lego
Slides 32–36
- Slides 33–34: the building blocks, then trace one request from phone to database. Reveal the three principles.
- Slide 35: 15-minute build. Teams build and save their architecture in ChaiCart Live. Captains circulate asking: “What happens if this box dies?”
- At 15 minutes, reveal the Diwali twist and start the 7-minute timer.
- Region judging (8 min): captains walk their six tables with the six Well-Architected questions from slide 36 and pick the region's best design.
- Main stage (8 min): the four region winners show their saved app design and explain it in 90 seconds each. Pick the overall best.
Your facilitator freezes the approved app design. Tomorrow some game cards will ask whether your design had autoscaling, multiple zones or a firewall. Your CTO will have to prove it.
Teams often forget multiple availability zones, a queue for bursts, caching and monitoring. Nudge with questions, not answers. Those gaps pay off on Day 2.
Break (15 min).
15:00 · 30 min
Microservices and the Human Kitchen
Slides 37–40
- Slide 38: one super-chef vs a station kitchen; reveal the cloud-native definition.
- Invite six volunteers from six different teams (at least one per region) to the stage. Everyone else counts out loud.
- Round 1, monolith (3 min): one volunteer does all five steps on each ticket. Count completed tickets.
- Round 2, microservices (3 min): five volunteers, one station each. Tickets flow down the line; each assigned volunteer confirms a handoff, which stores the server timestamp.
- Round 3, chaos (3 min): tap the Payment volunteer; they freeze. After 30 seconds, ask the room: “Where is order CC-007 right now?” Let the volunteers read the timestamps aloud.
- Slide 40: reveal the three insights; land the last card: the ticket was a distributed trace. If time allows, place an order in the deployed app and expand its trace in the facilitator console: cart → payment → pool → demo database → gateway → business events. These are logical services in one Node process, not independently deployed microservices.
Pick energetic volunteers and give them the mic. The chaos round is the key moment; don't skip it.
15:30 · 25 min
Industry use cases: Gallery Walk
Slides 41–42
- Assign two teams to each station (a region's six teams start at six neighbouring stations). Every 3 minutes, all teams move clockwise to the next station.
- Six rotations cover all six case studies. Each team adds digital reflections: Which cloud concept saved them? and What would ChaiCart copy?
- While teams walk, captains read notes and flag the best ones. Read out four and award them.
- Link the Capital One poster (breach and shared responsibility) to Day 2's security topic.
Moving 120 people takes time. Play music during rotations and say “move” clearly on the timer. If running late, stop after four rotations.
15:55 · 35 min
Observability intro, Dynatrace peek, quiz, cliffhanger
Slides 43–47
- Slide 44: the doctor analogy; then the monitoring vs observability quote.
- Slide 45, 5-minute Dynatrace peek: service map, a response-time chart, your story trace (“this is the order ticket from the kitchen!”), the Problems page. Ask the audience question on the slide.
- Slide 46, quiz: Open the Day 1 quiz from Run. COO submits one team answer per question; lock before advancing and use the quiz panel to score +10 per correct.
- Slide 47: Preview the incident teaser; evidence unlocks in the app on Day 2.
- Show the Day 1 leaderboard in presentation mode.
After Day 1
Wrap-up
- Freeze approved architecture designs for all 24 teams in Workshop.
- Delete
rg-chaicart-day1. Export the leaderboard CSV as a backup.
- Check digital incident evidence and keep its solution protected.
Day 2 · Cloud Operations, Observability & Business Systems
Keep ChaiCart Alive
09:30 · 10 min
Agenda, X icebreaker, warm-up and Breaking News
Slides 1–5
- Slide 1: use the build–test–recover illustration to connect yesterday’s design to today’s incident work. The loop is observe → explain → restore → learn. The Azure demo simulates dependencies; student designs do not provision cloud resources.
- Slide 2: walk through the Day 2 agenda, breaks and final Demo Day.
- Slide 3: one team per region shouts one word for Day 1. Show the leaderboard; remind everyone of the five roles.
- Slide 4: X learning circle (5 min kickoff, continuing throughout Day 2). Students improve their bio and find peers, then share workshop photos or takeaways with #ChaiCartCloudWorkshop. Show the hashtag search and announce the Best Workshop Post or Photo award. Collect best-post links through captains before the final awards; judge learning value, creativity and workshop spirit, not likes. Ask permission before sharing someone’s photo; no public post is required. Captains approve +100 reviews after all five members complete; the captain or facilitator applies each review once; posting is optional.
- Slide 5: read the breaking news dramatically; reveal the “payment stuck on spinner” line and pause.
09:40 · 15 min
Performance, reliability and Guess the Downtime
Slides 6–9
- Slide 7: the four user-facing signals. Explain p95 with the “95 happy, 5 tweeting” example.
- Slide 8: Open Guess the Downtime; COO submits four annual-minute guesses. Captains compare the closest team per row.
- Ask: “Does a chai app need five nines? Does UPI?”
- Slide 9: the six reliability patterns, linked back to the Human Kitchen.
09:55 · 15 min
Monitoring vs observability, APM, golden signals
Slides 10–12
- Slide 11: smoke alarm vs fire investigator; reveal APM.
- Slide 12: the golden signals, RED and USE. Stress the final hint about connection pools. It is a deliberate clue for the mystery.
“You have the theory. Now you have a crime to solve.”
10:10 · 50 min
Murder mystery: Who Killed Checkout?
Slides 13–17
- Slide 13: Open the digital mystery evidence from Run. Captains inspect stored timestamped accusations.
- Slide 14: introduce the six suspects with some theatre.
- Slide 15: rules and scoring. 25-minute timer. COOs submit one final accusation in Live; the app records the server timestamp and locks it. No paper forms or handwritten times are needed.
- Hints (announce to the whole room if most teams are stuck):
- At 10 min: “Compare database connections before and after 08:58.”
- At 15 min: “Look at the log lines from two days ago, then today's.”
- At 20 min: “In the failed trace, what's missing?”
- While teams work, captains review the evidence and proposed fixes. In the facilitator mystery panel, tick good fix/prevention submissions before applying. The grader uses server timestamps to add the +100 first-correct killer-and-weapon bonus; do not award that bonus again manually. Lock, reveal the solution on the slides, then press Apply mystery credits once.
- Slide 16, reveal: go suspect by suspect, innocents first. Ask the room to shout “innocent” or “guilty” before each reveal.
- Slide 17, lessons: stress the irony: a cost-saving change caused the outage. After the reveal, use the deployed app companion if rehearsed: baseline in Healthy / Roll back, select Deploy Dave: pool 50 → 5, launch the surge, compare waits/504s and missing gateway spans, then roll back and stop traffic. Keep controls hidden during the incident investigation. Live timings differ from prepared evidence.
The solution in one breath
Deploy Dave (payment-service v2.3.1 at 08:58) cut the DB connection pool from 50 to 5 to save cost. Traffic Tara (3× launch traffic at 09:00) exhausted the five connections. Requests waited 30 s and timed out, and checkout returned 504s. The database, gateway, cache and disk are innocent: the database was starved, not overloaded.
Most teams first accuse Postgres Pete or Traffic Tara. That is the teaching moment: “Traffic was the trigger, but what made the system fragile?” Accept “the 08:58 deployment” as the killer; award weapon points only for the connection-pool explanation.
Break (15 min). “After the break you'll do the same thing in a real tool with real data.”
11:15 · 45 min
Logs, metrics, traces and the Treasure Hunt
Slides 18–22
- Slides 19–20: the same incident as a metric, a log and a trace; how tracing works (trace ID = ticket number, span = station).
- Slide 21, Part A, Dynatrace (~20 min): you drive the screen. Read each question, give teams 90 seconds to discuss, then take a suggestion from a random table on where to click. Teams write answers from what they see.
- Part B, Azure (~15 min): share the deployed ChaiCart QR code. Teams open the menu and sign in with Google to place demo orders. Show the App Service resource and the app console. Request
/chai-not-found for a real 404. Show Live Metrics and run the worksheet KQL only after Application Insights ingestion is verified; otherwise use prepared evidence or a separately instrumented app. Confirm the budget alert exists before presenting Q14; budget alerts do not cap spending.
- Discuss answers; captains review the stored 15 responses and evidence before proposing credits.
- Slide 22: recap with the DQL examples: observability is asking new questions without shipping new code.
120 phones on one Wi-Fi can be slow. Mobile data works too; tell students they can use it. If the environment has no failures, switch Q1 and Q3 from “failed” to “slowest”.
12:00 · 40 min
SRE, SLOs and Error Budget Poker
Slides 23–26
- Slides 24–25: the Treynor Sloss quote, then SLI → SLO → SLA. Do the error-budget maths live: 30 × 24 × 60 × 0.1% ≈ 43 minutes.
- Slide 26: each SRE holds the 43-minute budget. Read one card aloud and start the 60-second timer. Teams discuss A/B; the COO submits the shared choice on their phone. Use Issue immutable staff rolls for this card when required, before submission. Next card advances; End the month enables Apply final credits. Apply once; do not collect separate paper choices or dice rolls.
- Cards 3 and 8 depend on the Day 1 architecture: the CTO must show proof on the frozen app design to the captain. Before Day 2, approve and freeze every completed team design, not only the four winners.
- Card 10 ends the month. Ask the team deepest in the red what it would do differently.
If time is short, play cards 1, 3, 5, 6 and 10. Keep card 6 (automate toil); it teaches the most.
Lunch. “After lunch we write the postmortem for this morning's murder.”
13:25 · 15 min
Blameless postmortem
Slide 27
- Explain the blame vs blameless examples on the slide; start the 12-minute timer.
- Captains circulate. If they hear “Dave messed up”, they ask: “What in our system let that change reach production?”
- Each captain picks the best postmortem in the region. One region winner reads its root cause and action items on stage; choose the overall best.
13:40 · 45 min
DevOps, Digital Delivery Factory, DORA, CI/CD demo
Slides 28–32
- Slide 29: the DevOps loop, CI, CD and IaC. Keep it short; the game makes the point.
- Slide 30: Open Digital Delivery Factory from Run. Start round 1 in Workshop for three minutes: COO processes six tickets serially. Stop and discuss queues. Start round 2 for three minutes: CEO plans, CTO builds, SRE tests, COO deploys concurrently. Failed tests return to Build. Compare successful deployments, frequency, lead time and correction time. Region winner: successful deployments, then fewer failures, then lower lead time; +50. These classroom metrics are DORA teaching proxies.
- Slide 31: DORA metrics and deployment strategies; land the Deploy Dave / canary joke.
- Slide 32, CI/CD demo: use the repository’s existing ChaiCart workflow. Until Azure deployment credentials are configured, demonstrate passing and failing tests on a rehearsal branch/PR and explain that deployment is skipped. Once deployment is enabled and rehearsed, change
chaicart-demo/public/index.html, show the successful deploy, and refresh the Azure URL. Use a controlled failing-test change and revert it immediately after the demonstration; do not leave main broken.
Rehearse with the four station roles. Keep round controls on the facilitator laptop; teams cannot extend deadlines.
14:25 · 35 min
Security, scalability, cost and Cloud Bill Shock
Slides 33–38
- Slide 34: shared responsibility; callback to the Capital One poster.
- Slides 35–37: move briskly. Ask one CFO per region which FinOps lever it would pull first.
- Slide 38: launch Cloud Bill Shock from Run. The app deals three cards per team; the CFO reads the costs and the CTO checks the frozen design. COO selects the preventing control and submits. The app computes design effects and half-refunds for correct controls; Lock, then Apply bill credits once.
- Ask each captain for the most painful card in the region, and the room: “Who has seen this happen in real life?”
Break (10 min). “Now let's leave the servers and look at the systems that actually run a business.”
15:10 · 35 min
Business systems and Follow the Order
Slides 39–43
- Slide 40: “Which of these have you used without knowing it?” (e.g. HCM when applying for an internship).
- Slide 41: Commerce starts with the order token, reads its line into the mic, confirms the digital handoff. Each system does the same.
- Twist: “ERP is down for maintenance!” The ERP volunteer sits. What still works (order, loyalty)? What piles up (invoices)? This is why queues and events matter between systems.
- Slides 42–43: integration, why SaaS wins, the role of cloud in digital business. In the deployed app, select Pause ERP, place an order, show the queued invoice alongside completed CRM/SCM/HCM/BI events, then select Healthy / Roll back to replay. This is simulated event processing; the app does not connect to SAP, Salesforce or other business systems.
15:45 · 30 min
Trends, Career Tarot, skills, hot-seat Q&A
Slides 44–48
- Slide 45: vote with hands for the trend that will matter most in five years.
- Slide 46: students browse the digital career cards in My workshop and discuss roles with teammates. Launch My 90-day cloud plan from Run; every student submits their own plan, not a shared COO answer. 10-minute timer.
- Slide 47: Preview the incident teaser in the deck; evidence unlocks in the app on Day 2.
- Slide 48, hot seat: reveal one question at a time and take two or three answers each via the handheld mic. Close with open Q&A about working in the industry.
This is the section students remember for their careers. Share real stories: an incident you worked on, what interviews look for, what a day in your job looks like.
16:15 · 45 min
Final quiz, Demo Day, awards, close
Slides 49–51
- Slide 49: Run the Day 2 quiz with the same team submission and scoring workflow; verify the final leaderboard.
- Slide 50, region round (15 min): in each quadrant, every CEO pitches for 90 seconds using the five points on the slide. The region votes for its finalist.
- Grand final (15 min): four finalists pitch for 3 minutes on stage (timer on slide). You and the captains choose Most Creative Pitch.
- Awards (5 min): show the final leaderboard in presentation mode. Announce Richest Startup, the four Region Champions, Sherlock, Best Architecture, Best Postmortem Most Creative Pitch and Best Workshop Post or Photo. Review the captain shortlists and announce one individual X winner.
- Slide 51: Share the cheat-sheet link, collect individual post-surveys in the app, record awards, verify attendance, mark the workshop complete, then publish permanent results and digital certificates in InspiredLearning.
You built it, you broke it, and you kept it alive. That's the job. Every cloud engineer in the world does exactly what you did today, just with more zeros on the bill.
Contingency
Running late? What to cut
| Behind by | Day 1 | Day 2 |
| 10 min | Skip Cloud Bingo | Guess the Downtime: reveal all rows at once |
| 20 min | Gallery Walk: 4 rotations only | Error Budget Poker: cards 1, 3, 5, 6, 10 |
| 30 min | Deployment debate: defend 4 scenarios, not 8 | Digital Factory: round 2 only, compare verbally |
| 45 min | Shark Tank: region round only, no grand final | Treasure Hunt: Dynatrace only (skip Azure) |
Never cut: the Human Kitchen chaos round, Architecture Lego, the murder mystery, and Demo Day.
| Situation | What to do |
| Wi-Fi fails | Keep downloaded decks and prepared evidence available. Pause Live submissions and scoring until connectivity returns; cached drafts are not confirmed answers. Use recordings/screenshots for demos. Do not switch to a second live leaderboard. |
| Fewer than 4 helpers | Assign each captain account one region. With two helpers, facilitator reviews the other two regions. With none, CEOs can coordinate discussion/votes, but do not acquire staff access; facilitator reviews and applies scores. |
| Absent students | Resolve team allocation before students register. Do not merge registered teams or reset roles casually. Networking +100 requires all five registered members to complete; handle attendance exceptions explicitly with the facilitator and a recorded reason. |
| Quiet room | Always let teams discuss before asking for an answer; reward the first volunteer generously. |
| One team dominates | Direct challenge questions to other teams; lean on region and judged awards. |
| Leaderboard closed | Reopen ChaiCart Live Console with the same session. Credits remain in its persisted ledger. Do not switch to the legacy static leaderboard or reset the session; export results when connected. |
Share this link with each region captain
Region captain briefing
Thank you for helping run the ChaiCart Cloud Workshop. 120 students work in 24 teams of five, grouped into four regions of six teams. You are the host for one region for both days.
Your region
Region: ______________________ Teams: __________-1a to __________-1f
Your job in one line
Keep your six teams moving, check stored answers, and save reviews for facilitator approval.
Activity by activity
Networking: LinkedIn at the Day 1 opening, GitHub after lunch on Day 1, and X at the Day 2 opening. Approve +100 per team per activity when all five teammates complete it; captain or facilitator applies the review once; existing accounts, connections and follows count. Record each only once; maximum +300. Ask the COO for completion, not account credentials or screenshots.
| Activity | What you do |
| Day 1 · Human Timeline | Compare correct locked digital submissions and server timestamps. Propose +50 for the first correct team in your region. |
| Day 1 · Service sort | Check stored service choices against the protected key (×5 credits). |
| Day 1 · Deployment debate | Assign one digital scenario per table. Note any team that wins a defence (+30). |
| Day 1 · Shark Tank | Use the six digital briefs and platform options. Run the region round: 1-minute pitches, each team votes (not for itself). Vote-panel winner +50, goes to the grand final. Facilitator awards from that panel; do not also propose duplicate pitch points. |
| Day 1 · Bingo | Check the digital board against called terms. Coordinate the first three valid room-wide claims; propose +30 only for confirmed winners. |
| Day 1 · Architecture | Ask “What if this box dies?” while teams build. Judge with the six questions on slide 36. Region best +50, goes to stage. |
| Day 1 · Gallery Walk | Read notes during rotations; flag the best one in your region to the facilitator. |
| Days 1–2 · Quizzes | Verify team submissions; the quiz panel applies +10 per correct answer. |
| Day 2 · Guess the Downtime | Compare stored guesses in your region for each row (+20). Save the total review (up to +80), then Award reviewed credits to team once. |
| Day 2 · Murder mystery | Review the timestamped accusations and protected rubric; flag the earliest correct team. |
| Day 2 · Treasure Hunt | Review all 15 answers and proof fields, up to 150. Save an approved review, then Award reviewed credits to team once. |
| Day 2 · Error Budget Poker | Check frozen design proof and help the SRE reason. The facilitator issues staff rolls and applies computed final credits from the Poker panel. |
| Day 2 · Postmortem | Ask “what in our system allowed this?” when you hear blame. Pick the region best (+40). Save the approved review and Award reviewed credits to team once. |
| Day 2 · Digital Factory | Compare round-2 successes, then fewer failures, then lower lead time; approve +50 for region best, then Award reviewed credits to team once. |
| Day 2 · Bill Shock | Help CFO/CTO discuss digitally dealt cards and frozen designs. COO selects controls; the app computes half-refunds. Facilitator applies bill credits. |
| Day 2 · Demo Day | Run the region round: 90-second pitches, region vote. Send your finalist to the stage. |
Keep your region's energy up: learn the team names, cheer your region's finalists, and bring any problems (missing materials, a team of three) to the facilitator during the next break.
Captain quick reference
Digital review and scoring
- Sign in at Captain dashboard, enter the session, and select one of your six teams.
- Read the stored response, protected rubric and roster. Save credits and feedback as an approved review.
- A captain can use Award reviewed credits to team; the facilitator can apply the same review once from Workshop. For other activities, bonuses or corrections, use the captain’s any-activity credits form with a specific reason; stay within your region and do not repeat automatic quiz awards. Students see feedback and the credit ledger in My workshop. Quizzes, votes, mystery, Poker and Bill Shock use their dedicated scoring controls.
- Networking earns +100 for each activity after all five members confirm completion: LinkedIn, GitHub and X. Maximum +300 per team; existing connections count. Never ask for passwords.
- Flag X entries for the best post/photo award; public posting is optional. Captains verify regional attendance in InspiredLearning; the facilitator publishes permanent certificates after award entry and workshop closure.
- Corrections use the immutable ledger and Undo compensation. Do not reset an active workshop leaderboard.
Closing and beyond
Permanent learner account · InspiredLearning
At closing, open Workshop staff, load the same session and import existing registrations if needed. Captains verify their region’s attendance; mark absentees not-attended. Finish scoring and awards, mark the workshop complete, then the facilitator publishes verified results. Students use their same Google account to download a certificate and see personal results, team standings, confidence changes and 100 learning points per completed workshop. Repeated publication does not duplicate points. Publish corrected results as a new version. Review all learners before deleting the Live session; permanent history survives deletion. Discord remains available after the workshop.
- Captains verify actual attendance within their own region.
- Facilitator closes the workshop after final scoring, then publishes verified results.
- Students sign in with the same Google account, review their records and download their PDF.
- For older sessions, import existing registrations first. Registration alone never awards completion or a certificate.
- Correct or revoke archived certificates from Permanent records; repeat publication never duplicates learning points.