Community for ops engineers
2nd AI SRE & AIOps Workshop: Pune Edition
Build or buy your AI SRE? Work it out, hands-on.
Spend one Friday working it out with up to 23 other SRE, platform and cloud ops engineers in Pune, before you sign a contract or write the first line of code.
You leave with your own call, scored against your own org, plus the cost numbers and guardrails to back it up.
Practitioners only · apply to get in · pay only after approval Download the flyer (PDF)
What you take back to work
Worth one FridayScored against your own org, ready to defend to leadership.
Cost estimate at your alert volume, and where token spend runs away.
Model-upgrade checklist plus patterns that stop hallucinations.
Written down on the day. We email it back to you on day 30.
Build or buy? Every ops team is asking.
Buying gets you speed. Building gets you control. The right call depends on four questions most teams haven't worked through yet:
- 01What does the operating brain need: memory, context, routing, harness?
- 02How do you move between model versions without breaking production?
- 03What does it cost to run, and where does token spend run away?
- 04How do you stop hallucinations once agents touch production?
You work through all four with your own numbers, and "build" is a perfectly good outcome.
You'll operate it, not watch it
Hands-on from your own seat- Every seat gets a live Kubernetes cluster. Accounts are handed out on the day. Nothing to set up beforehand.
- Four AI models go head to head on the same ops problem.
- The agent runs with memory on, then off. You watch the answer change.
- The room red-teams the agent. One goal: break it before it breaks production.
- Real production blockers get worked on. Bring what bit you. The group digs in.
Six topics on the table
Real production stories, no slidesEveryone says "agent". Which one do you actually need to build?
Most ops agents fail in the setup, before their first answer.
It solved this exact incident last Tuesday. Today it has no idea.
Guardrails, permissions and blast-radius control before prod.
Wrong 5% of the time in production means switched off in a week.
When the agent costs more than the hours it saves.
What you do on the day
Fri 9 Oct · 10 AM to 2 PM- 10:00 AMChai and arrivals
- 10:15 AMYour #1 production blocker, on the table
The day is built around the room's blockers.
- 10:30 AMBuy vs build
Score your own org against a decision framework and leave with a verdict.
- 11:15 AMThe operating brain
Memory, context, model routing and the harness, built live on your own cluster.
- 12:30 PMLunch with the room
- 1:00 PMWhat it costs
Price your own architecture at your alert volume and see where token spend runs away.
- 1:25 PMBreak the agent
Red-team it yourself, scored on a live leaderboard. Failure stories stay in the room under Chatham House rule.
- 1:55 PMYour 30-day commitment
One change, written down. We email it back to you on day 30.
24 seats, 2 per company. Apply now, pay only after approval.
Apply for a seatLeading a team? Nominate 2.
Send the one or two engineers who will own AI in your ops stack. It's 2 seats per company, so pick the pair who will act on it. They come back with the four things above, for ₹500 each and one Friday, 10 AM to 2 PM. Share this page with them. Each person applies separately.
I'd like to attend the AI SRE & AIOps Workshop in Pune on Fri 9 Oct, 10 AM to 2 PM. It's hands-on and capped at 24 people. I'll come back with a build vs buy verdict for our team, a cost estimate at our alert volume and one change we can make in the next 30 days. It costs Rs 500. Details: https://luma.com/53gmriwhEmail it to your manager
Who should apply
- SREs, platform engineers, DevOps and cloud ops engineers who work hands-on with production systems
- Leads and managers who own reliability or cloud spend and have to decide what AI does in their ops stack
- Teams at any stage: not started, evaluating, piloting or already running agents in production
- Anyone who doesn't work on production systems. That's why the room is capped.
How to get in
- 1Apply on Luma. Five short questions.
- 2We read every application. You hear back within 1 hour.
- 3If you're approved, you get an email with a link to pay ₹500. Check your spam folder for it.
- 4Seats go to the first 24 approved people who pay.
- Bring a laptop with a browser.
- Know your team's rough monthly alert volume. You'll use it in the cost session.
Edition 1 recap: Hyderabad, 18 Sep 2026
25 registered and 24 engineers attended, hands-on from 10 AM to 2:30 PM.
Questions
Is this a sales pitch?+
No. GoGoAI Dev, a community for ops engineers, organises it. NudgeBee, an AIOps company from Pune, is the partner. You score your own org, test four models yourself and red-team the agent, so the verdict is yours. "Build" is a valid answer.
What does it cost, and when do I pay?+
₹500 per person, lunch and chai included. You pay only after your application is approved, through the link in the approval email.
Do I get a receipt?+
Yes. Razorpay emails a payment receipt after you pay.
Can a colleague take my seat?+
Yes, if they apply and we approve them. Email devgogoai@gmail.com.
Is it recorded?+
No. The failure stories are shared under Chatham House rule.
Where exactly?+
SB Road / Shivaji Nagar, Pune, easy to reach for everyone. Exact venue shared with approved guests before the event.