Urmish Thakker
Senior Engineering Manager - Applied NLP at SambaNova Systems
California, United States
Summary
👤
Senior🎓
Top SchoolUrmish Thakker is a Senior Engineering Manager specializing in Applied NLP with 11 years of experience building and scaling high-performance ML teams and systems from research to product. At SambaNova he leads NLP efforts across LLM ecosystems, speech products, and HW–SW co-design, blending deep systems knowledge with product delivery. His background at Arm and TI spans efficient DNN execution, quantization, pruning, and low-power accelerator design, with publications and patents bridging academia and industry. A founding contributor to MLCommons’ TinyML Perf working group, he helped define benchmarks now used by industry and academia to validate MCU-class NN accelerators. He contributes to open-source NLP tooling—fixing subtle rendering and template bugs in the prominent PromptSource repo—demonstrating attention to both model and tooling quality. Based in California, he combines technical depth in ML systems with proven hiring and team-scaling experience (35+), often translating research innovations into deployable engineering outcomes.
11 years of coding experience
7 years of employment as a software developer
Master of Science - MS, Computer Science, Master of Science - MS, Computer Science at University of Wisconsin-MadisonB.E., Electrical and Electronics Engineering, 9.30/10, B.E., Electrical and Electronics Engineering, 9.30/10 at Birla Institute of Technology and Science
English, Gujarati, Hindi
Stackoverflow
41reputation
3kreached
0answers
1question
Github Skills (11)
python10
natural-language-processing9
streamlit9
machine-learning8
jinja28
arm6
opencv6
cmake6
linux6
ffmpeg6
javascript5
Programming languages (4)
Github contributions (5)
bigscience-workshop/promptsource
May 2021 - Apr 2022
Toolkit for creating, sharing and using natural language prompts.
Role in this project:
Full-stack Developer
Full-stack DeveloperContributions:3 reviews, 18 commits, 15 PRs in 11 months
Contributions summary:Urmish primarily addressed issues related to hyphen rendering within the promptsource application. Their contributions involved modifying the codebase to replace hyphens with underscores in several areas, including template jinja, example keys, and dataset column names, to prevent rendering issues. They also refactored code to use a function instead of a direct dictionary manipulation and resolved merge conflicts. Furthermore, the user implemented and refined dataset filtering logic and template count retrieval functionality.
nlppromptspythontransfer-learningsharing
Urmish/MLPerf_BERT
Apr 2021 - Sep 2023
Contributions:1 PR, 9 pushes, 3 branches in 2 years 6 months
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.