Dipika Sikka
Machine Learning Technical Lead at Red Hat
Boston, Massachusetts, United States
Summary
🤩
Rockstar🎓
Top SchoolDipika Sikka is a Machine Learning Technical Lead with nine years of experience applying computer vision and ML to healthcare and production-grade systems, now leading ML efforts at Red Hat after Neural Magic’s acquisition. Trained in biomedical engineering and an MS from Columbia, she blends domain knowledge in health sciences with deep technical chops in model optimization and inference. Her work includes significant open-source contributions to memory- and compute-efficient LLM serving (vllm), focusing on low-bit and dynamic per-token quantization to reduce footprint without sacrificing throughput. She has a track record of moving research prototypes into industry products across startups and research labs, from biomedical imaging at UCLA to AI at Covera Health and VantAI. Based in Boston, she brings a pragmatic engineering style that prioritizes deployable performance and measurable clinical impact. A less obvious strength: she pairs hands-on firmware/sensor prototyping experience from early roles with large-scale ML systems expertise, enabling cross-stack solutions.
9 years of coding experience
7 years of employment as a software developer
Ontario Secondary School Diploma, Ontario Secondary School Diploma at Stephen Lewis Secondary SchoolMaster of Science - MS Engineering, Master of Science - MS Engineering at Columbia UniversityBachelor of Applied Science (B.A.Sc.) Biomedical/Medical Engineering (Co-op), Bachelor of Applied Science (B.A.Sc.) Biomedical/Medical Engineering (Co-op) at University of Waterloo
English, Hindi, French
Github Skills (8)
cuda10
quantization10
pytorch10
transformer10
inference10
llm10
deepspeed9
roc6
Programming languages (3)
Github contributions (5)
vllm-project/vllm
Apr 2024 - Mar 2025
A high-throughput and memory-efficient inference and serving engine for LLMs
Role in this project:
ML Engineer
ML EngineerContributions:147 reviews, 75 PRs, 213 comments in 10 months
Contributions summary:Varun Sundar Rabindranath's contributions primarily revolve around integrating and optimizing quantization techniques for large language models within the vllm project. He implemented support for activation quantization and dynamic per-token quantization, specifically focusing on 4-bit and 8-bit quantization schemes. His work also includes adapting the project to utilize the new compressed-tensors config and integrating support for various quantization formats, demonstrating a focus on memory efficiency and performance optimization for LLMs.
amdcudadeepseekgpthpu
neuralmagic/compressed-tensors
Apr 2024 - Apr 2025
A safetensors extension to efficiently store sparse quantized tensors on disk
Contributions:141 reviews, 88 PRs, 185 pushes in 11 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.