Dipankar Sarkar

Dipankar Sarkar

Infrastructure for the AI-native world

About

I build the infrastructure for an AI-native world: the runtime, safety, performance, and tooling layers that let AI agents actually ship. Across neul-labs (production agent infrastructure and dev tools), Skelf-Research (the experimental edge), and cognisoc (local and edge LLM inference), I work on things like Rust drop-in speedups for the AI stack, safe data access and governance for agents, and running LLMs anywhere. Earlier I was a Principal Engineering Consultant at Nykaa and an ML Consultant at Hike, published at ML conferences, and filed 60+ patents. Here to support other builders, give honest product feedback, and share what I learn shipping the agent stack. Always happy to talk shop.

Badges

2

Top 5 Launch
Top 5 Launch
Tastemaker
Tastemaker
Veteran
Veteran
Gone streaking
Gone streaking

Maker History

Forums

2yr ago

What Generative AI - Your open-source guide to mastering GenAI in business

Comprehensive, open-source playbook for implementing Generative AI in business. From use case identification to ethical considerations and culture change.

2yr ago

Fragaria : CoT + RL = Reasoning - Self-improving AI reasoning engine for developers

Fragaria is an open-source AI reasoning API that combines Chain of Thought with Reinforcement Learning. It tackles complex problems, learns from each interaction and easily integrates with OpenAI, Groq, and Together.ai. Perfect for researchers and developers.
View more