Your AI agent can call an API, touch your data, and execute a workflow. Who said it could?

Enterprise AI has moved past “get the model into production.” The engineering problem still remains: “how do I control what this thing is allowed to do.”

📅 Schedule - https://hasgeek.com/rootconf/platform-engineering-meet-up-september-5/
🕑 Saturday, 5 September, 3 PM – 5 PM
📍 Venue: Blinkit office, Board Room, 7th floor, Vaishnavi BVS Senate, Bannerghatta Road
🔴 Live stream available exclusively for Rootconf members — free in-person seats for everyone else.

Platform Engineering meet-ups are curated by:

• Talina Shrotriya, Senior Engineering Manager at Couchbase
• Vivek Pemavat, Principal Engineer at Acceldata
• Bharadwaj Embar, Engineering Manager at Nutanix

Edition 6: Platform Engineering for enterprise AI — security, guardrails, governance, sandboxing & observability

AI agents call APIs, touch enterprise data, execute workflows, and make decisions on the user’s behalf. The questions we can’t stop mulling over are:

• How do you sandbox agent execution?
• How do you enforce policy consistently across models, agents, and tools?
• How do you audit what an agent actually did, and why?
• How do you make all of this reusable instead of rebuilding it for every AI app?

Featured talks

1. Beyond AI sandboxing: decoupling AI from execution — Gowtham Sadasivam (Senior Staff Engineer, Acceldata) makes the case for decoupling AI agents from execution entirely: instead of sandboxing the agent, run it against a secure remote dev environment on infrastructure you control, over MCP, with any AI coding assistant.

2. Observing and governing agents you did not build — Vivek Chand (AI Engineer, Booking.com; creator of ClawMetry) on why the tools already log what they do, but every vendor exposes it differently — and how to read those formats yourself, land everything in one queryable store, and put a real policy enforcement point (spend caps, loop detection, a kill switch) on agents you don’t control the source of.

🎤 Birds of a Feather

The new age of building on object storage

Why are database and database-adjacent systems increasingly being built directly on S3? A growing list of systems are betting on object storage as the primary layer, not just cold storage.

Open discussion:

• Why build on S3 in the first place — what does it actually buy you?
• Why not — where does this break down?
• What kinds of systems does this pattern not make sense for?

Bring your opinions & experiences.

Editors’ note

Our CFP follows our published conflict-of-interest policy. As usual, any editor with a conflict recuses themselves from the review and decision.

🎬 Catch up on Edition 5

The crowd favourite — Parth Agrawal’s talk on running Apache Druid at 700 MB/s and building a platform that mostly runs itself → https://lnkd.in/dqmSFn_A

💡 Why you should attend

• Hear real-world lessons from engineers securing and governing agentic systems in production.
• Connect with peers working on AI platform security, sandboxing, and observability.
• Walk away with patterns and blueprints for reusable AI guardrails — not one-off fixes.

🧭 How to attend

• In-person participation is free — just click Register to confirm your seat.
• Remote participation available for Rootconf members. 👉 Get your membership from https://hasgeek.com/rootconf#memberships

🏢 Hosted by Blinkit

Thanks to Blinkit for hosting Edition 6 — and to Sumanth Reddy, part of the Rootconf community, for making it happen.

Want to host a future edition? It gets your org tech branding and a room full of platform engineers — especially handy if you’re hiring. 📧 sales@hasgeek.com

📞 Contact

💬 Post your questions — https://hasgeek.com/rootconf/platform-engineering-meet-up-september-5/comments
📧 Email: info@hasgeek.com
📱 Call/WhatsApp: +91 7676332020