Karim El-refai
EECS C106A Head TA at Berkeley Artificial Intelligence Research
United States
Summary
🤩
Rockstar🎓
Top SchoolKarim El-refai is a software engineer and researcher with 10 years of experience bridging robotics, ML, and backend systems from UC Berkeley and BAIR. Currently Head TA for EECS C106A and a Berkeley AI Research assistant, he works on real-time mapping and deformable object manipulation while mentoring large undergraduate cohorts. His industry internships span autonomous vehicles and cybersecurity, and he built perception stacks for 170mph autonomous racing using multi-sensor fusion and EKFs. An active open-source contributor, he improved DSPy’s Pinecone integration and added local embedding support—highlighting a focus on robust MLOps and production-ready retrieval systems. Karim combines hands-on C++/ML engineering with curriculum development and a knack for making research tools deployable in real environments.
10 years of coding experience
2 years of employment as a software developer
Master of Science - MS, Electrical Engineering and Computer Sciences, Master of Science - MS, Electrical Engineering and Computer Sciences at University of California, Berkeley
Github Skills (11)
pinecone10
machine-learning10
vector-database10
backend10
dsym10
back-end-development10
python10
nlp9
mlops9
huggingface-transformers9
openai-api8
Programming languages (7)
Github contributions (5)
stanfordnlp/dspy
Dec 2023 - Dec 2024
DSPy: The framework for programming—not prompting—language models
Role in this project:
Back-end Developer & MLOps Engineer
Back-end Developer & MLOps EngineerContributions:2 PRs, 5 comments in 1 year
Contributions summary:Karim primarily contributed to the `pinecone_rm.py` file, focusing on integrating the DSPy framework with Pinecone vector databases. Their work involved fixing the passage format, addressing issues with scoring, and creating functionality to create the PineconeDB if it doesn't already exist. Additionally, they added support for local embedding models, likely to enhance the flexibility and accessibility of the retrieval process. These changes suggest a focus on backend functionality, infrastructure setup, and the integration of ML models.
nlpbertknowledgepredictlanguage-models
matteoguarrera/planning
Apr 2023 - May 2023
Diffusion Planning CS 282
Contributions:4 PRs, 10 pushes, 4 branches in 7 days
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.