Bharat Venkitesh

Senior Member of Technical Staff @Cohere

Toronto, ON, CA

Over 5+ years, I’ve led initiatives across the full stack of LLM and multimodal development—from distributed training and inference optimization to cutting-edge research and technical leadership.Key Contributions:2024–Present | Multimodal AI – Senior engineer leading key initiatives in vision‑language models (Command A Vision) and advancing multimodal foundation models spanning text, vision, and audio.2023–2024 | LLM Research & Scaling – Core contributor to Command R, Command R+, and Command A; worked on scaling laws, μP, long‑context models, and architecture design.2021–2023 | Inference & Serving – Built Cohere’s custom inference stack, achieving 4–8× throughput gains via CUDA kernels, quantization, and C++ optimizations.2020–2021 | Pretraining Infrastructure – Scaled distributed training systems and model parallelism for Cohere’s first large LLMs.Impact: Core engineer behind three generations of Cohere’s flagship LLMs, major inference cost reductions, and technical leadership driving next‑gen multimodal research to production.

University of Waterloo

Master of Science - MS, Electrical and Computer Engineering

Indian Institute of Technology Hyderabad

Bachelor's Degree, Electrical Engineering

Indian Institute of Technology Hyderabad

Bachelor's degree- Minors, Computer Science