Bhargav started writing code in 2009 at K.J. Somaiya College of Engineering in Mumbai. He spent the next decade crossing three continents — a Master's at Rochester Institute of Technology, software engineering stints at Intuit in San Diego and Appian in Virginia — before landing in Switzerland in 2021.

At Abacus Research he spent two years in the deep end: fine-tuning frontier models — Mistral, LLaMA3, Mixtral 8x22b — using LoRA, QLoRA and Unsloth. Not prompting wrappers. Actual weight modification, quantisation, inference optimisation with vLLM, SGLang and LMDeploy. He took accuracy from 70% to 90%. Then he built multimodal document extraction systems that read invoices the way humans can't — at scale, without getting tired.

Now at LGT Private Banking in Zürich — one of the largest private banks in the world — he leads the development of a GenAI enterprise search platform on MCP server architecture, Qdrant hybrid retrieval, and OpenAI-compatible endpoints. Every design decision has to survive a compliance review. Every answer has to be grounded and traceable. There is no margin for hallucination when the client is a family office with $10 billion under management.

Three master's degrees. Two published papers. Four languages. One problem he keeps solving everywhere he goes: the knowledge is there — nobody can reach it.