CS student at VIT and data science intern at Blend360, where I build LLM pipelines, agentic workflows, and decision intelligence systems for enterprise clients. I write about the things I notice along the way.
I build LLM pipelines, retrieval systems, and agentic workflows — mostly in Python, mostly for enterprise use cases. At Blend360 I built a persona generation agent that synthesised financial customer personas at scale, a decision intelligence dashboard for automated narrative reporting, and inference workflows running 100,000+ concurrent runs.
Outside of work I contribute to open source ML tooling, maintain an open-source JAX/Flax LLM repository, and write about what I find interesting in the AI agent ecosystem. I'm also working on a biomedical knowledge graph pipeline using TransE embeddings for drug-disease link prediction.