AI Research Consultant
Computer vision, robot learning, and edge AI — from research question to deployed system.
I take on the AI problems that sit between a research paper and a product: perception that has to work on real footage, robot policies trained on real demonstrations, and models that have to run on the hardware you already own.
I have consulted independently since 2017 — part-time while completing my PhD, full-time since. Most of that work has been with startups, getting computer vision off the benchmark and onto real hardware: robotics platforms on one side, CCTV and surveillance networks on the other, both under the same edge constraints. Doctoral and postdoctoral research behind it, and hands-on delivery as co-founder of Witsense AI — so the recommendation you get is one I have shipped myself.
Four practice areas. Most engagements combine two or three of them.
Feasibility & scoping
Short, focused review: is this tractable with current methods, what data would it need, and what would it cost to find out? Ends with a written recommendation you can act on.
Prototype to production
Hands-on build of the perception, policy, or deployment layer — with the evaluation protocol and failure analysis that tells you when it is genuinely ready.
Embedded R&D advisory
Ongoing technical partnership with your team: architecture review, hiring input, and training so the capability stays in-house when the engagement ends.
Research, applied AI systems, and startup consulting engagements. Click any card for details, demos, and code. Demos on client engagements are prototype or reduced MVP builds that illustrate the approach, never the client's production system.
Peer-reviewed work on few-shot learning, domain adaptation, and the cognitive evaluation of learning agents — with code where it exists.
Notes on applied AI, robot learning, and the gap between benchmarks and deployment.
9 years at D. Y. Patil College of Engineering, Pune · NBA/OBE accredited programme · Board of Studies Advisory Member (ongoing)
I take on consulting engagements in computer vision, robotics, and edge AI deployment — alongside collaborative research and speaking. If you have a problem that needs both research judgement and someone willing to write the code, get in touch.