I am a Founding Engineer #6 at LlamaIndex (Oct 2023 – Jan 2025).
I’m an AI Engineer based in San Francisco, focused on building and improving production‑grade LLM systems. My career spans applied AI, retrieval‑augmented generation, fine‑tuning, evaluation, and data‑centric AI, with roles at Mistral AI, LlamaIndex, and independent projects like Navarasa and BRAG.
Implemented techniques and papers including GraphRAG, Corrective RAG, Adaptive RAG, Mixture of Agents, RAFT using LlamaIndex.
At LlamaIndex I contributed advanced RAG features, authored an O’Reilly course on building RAG applications, and engaged the community through open‑source contributions and technical blogging.
My independent project Navarasa is a multilingual LLM fine‑tuned for 15 Indian languages, showcased at Google I/O and ranked among the top Indic models by Microsoft Research India.
The BRAG family comprises ultra‑cheap RAG‑optimized models trained for under $25 each using Llama 3.1 and Qwen2 Instruct.
BRAG outperformed larger models Cohere Command R+ and Llama 3.1 on NVIDIA’s ChatRAGBench.
I hold a Master’s in Data Science from IIIT Bangalore, with a focus on NLP, computer vision, and deep learning.
Current interests include LLM evaluation, synthetic data generation, reinforcement‑learning environments, and feedback‑driven model improvement. You can find me on GitHub, X/Twitter, and Medium.