After 15+ years in banking engineering and delivery transformation, I realized the biggest blocker to enterprise AI adoption is hallucination risk and data privacy. I am now building MemStrata to solve this. It's a local-first context engine that uses HRR-based symbolic triple memory to strictly verify facts before injecting them into an LLM's context window. I focus heavily on local orchestration, squeezing performance out of consumer GPUs, and building autonomous agent pipelines.

Always happy to talk about local-first AI architecture.

Email: yadu989@gmail.com