Usvsn Prashanth is an applied scientist with six years of experience designing and scaling large language models and interpretability research, currently at Microsoft. He has driven projects from ideation to deployment—optimizing training for 70B+ parameter models with tensor and data parallelism and reducing inference time by over 30%. His research at EleutherAI produced a semantic memorization taxonomy and scaling laws for memorization across time and model scale, reflecting a strong focus on model transparency and safety. He’s skilled at prompt-tuning, synthetic dataset generation, and parallelized training workflows that bridge research and production. Based in Hyderabad, he combines hands-on engineering with teaching and documentation experience, having authored transformer-from-scratch materials for Scaler Academy. An unusual strength: he frames experiments as system-level “spacecraft” builds—stacking training runs and tests until models reliably operate on their own.
6 years of coding experience
3 years of employment as a software developer
Bachelor of Engineering - BE, Information Technology, Bachelor of Engineering - BE, Information Technology at Matrusri Engineering college
Provides a minimal implementation to extract FLAN datasets for further processing
Contributions:8 commits, 1 PR, 8 pushes in 11 months
datasetsdata-processingdata-analysisdataset
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.
Request Free Trial
Usvsn Prashanth - Applied Scientist Ll at Microsoft