Aditya Jindal

I build the layer between "the model said something" and "production users got a useful answer that didn't blow the budget." Full-stack platform and AI infra — that seam where systems meet.

Three years in production, currently at Simbian.AI on the integrations platform - I'd rather be where the systems are getting built than where they're being maintained. Before that: Wells Fargo (RAG + RPA at enterprise scale), Maya Labs (YC S22) on the program-synthesis platform and Flatland augmentations, and Solvio.AI building math-OCR models with TensorRT/Triton. Earlier still: real-time embedded vision on Jetson hardware.

I'm most useful when a problem cuts across systems - when the "bug" is actually a query plan interacting with a connection pool interacting with a rate limiter.

Outside of paid work, I build what I or the people around me need. GameTools because EA quit writing tools for the games we play. A self-hosted media player because no one's Plex was going to scan my disk the way I wanted. Both still ship. Right now I'm folding agents into GameTools - converting the tools into MCP servers with gRPC dispatch, writing packet-analysis skills so an agent can reason about Blaze traffic to debug new packet shapes, and building the agent loop on top.

outside of work

1000+ hours in Battlefield and Factorio. The Battlefield hours aren't a coincidence - they're why GameTools exists. Here's a clip of me being mediocre at BF1, for honesty:

▸ gameplay-bf1.mp4youtube ↗

resume

You can download my latest resume.pdf or view my career credentials below.