Collaborate
Three narrow lanes
I'm not looking for a full-time role, and I'm not selling consulting. I am open to a small number of technical collaborations, selective advisory work, speaking, and research conversations. Narrow is the point — these are the three places I'm actually useful.
Researchers and builders
If you're working on dependable AI — evaluation, verification, calibration, observability — or on small and local models, edge inference, or geospatial and physical intelligence, I want to compare notes. Especially if you have results that contradict mine.
Founders and technical leaders
Selective technical sparring and advisory around AI systems architecture, data and knowledge architecture, evaluation and reliability, and what it actually takes to put a probabilistic system into a real workflow. I've watched agents fail for organizational reasons more often than model reasons, and that's usually the conversation worth having.
Universities and communities
Applied research labs, technical talks, and student programs. I run the Acropolis Edge Lab in Indore on this model — phone-only projects, run as experiments rather than demos — and I'm happy to help someone else start one.
How to start
The most useful opening is a specific question or a system you're stuck on, not an introduction. If you want context first, the research questions and the writing are the honest version of where I am.