About

I’m Prathmesh — a computer vision researcher and machine learning leader with over ten years spanning peer-reviewed research and production systems.

I’m Senior Vice President of Machine Learning at Infocusp Innovations, where I lead the computer vision group across concurrent LLM and computer vision programmes. Recent work includes TensorRT-accelerated visual SLAM for aerial platforms, agentic systems with LLM-as-a-judge evaluation pipelines, hybrid multimodal retrieval, and defect detection on production lines.

I hold a PhD from the Chair of Pattern Recognition at FAU Erlangen-Nürnberg. My dissertation addressed scene understanding in digital humanities — art history, classical archaeology, and Christian archaeology — extending object detection, pose estimation, and compositional analysis to artwork. That work produced ICC and ICC++, explainable feature learning methods for understanding image composition.

The through-line across both halves of my work: I take on problems where standard models fail — non-standard imagery, scarce labels, tight compute budgets, and outputs that have to be auditable.

Where to go next

• Selected work — what I’ve built, and what it measurably did
• Research — publications, talks, and teaching

Outside of that, I’m fascinated by mathematics and drawn to problems built on it. I also write occasionally about work, focus, and how this field keeps changing.

Cogito, ergo sum. — René Descartes

If you’d like to start a conversation, message me on X or LinkedIn.