Erik Nijkamp
Research Director at Salesforce
San Francisco, California, United States
Summary
🤩
Rockstar🎓
Top SchoolErik Nijkamp is a Research Director and ML scientist with 15 years of experience building and scaling large generative models, currently leading R&D on agentic memory systems and continual learning at Salesforce in San Francisco. He created influential open-source families—CodeGen for program synthesis, ProGen2 for protein language modeling, and XGen for conversational LLMs—and played a hands-on role in their architectures and training pipelines (including Deepspeed/TPU-v4 work). His career bridges deep academic grounding (PhD/MS in Statistics from UCLA) with startup and product experience, having founded and exited a deep-tech QA company before driving enterprise GenAI adoption. Notably, his CodeGen work competed with OpenAI Codex, reflecting a rare combination of research rigor and production-grade model delivery.
15 years of coding experience
16 years of employment as a software developer
Master of Science (M.Sc.) Computer Science, Master of Science (M.Sc.) Computer Science at Technische Universität BerlinUniversity of California, Los Angeles
Stackoverflow
1reputation
0reached
1answer
0questions
Github Skills (23)
pytorch10
language-model10
apache-flink10
python10
evaluation10
memory-management10
big-data10
machine-learning10
flink-sql10
java10
javas10
eval10
generative-model10
tokenize9
scala9
Programming languages (4)
Github contributions (5)
salesforce/CodeGen
Mar 2022 - Dec 2022
CodeGen is a family of open-source model for program synthesis. Trained on TPU-v4. Competitive with OpenAI Codex.
Role in this project:
ML Engineer
ML EngineerContributions:29 commits, 5 PRs, 50 pushes in 8 months
Contributions summary:Erik contributed to the development and maintenance of the `salesforce/codegen` repository, which focuses on code synthesis. Their commits include the initial import of core modeling files, specifically `modeling_codegen.py`, indicating a role in defining and implementing the underlying model architecture. Further contributions finalize checkpoints and add new model configurations, solidifying their work in model development and potentially model training/evaluation within the project. They also added deepspeed configuration for training.
openai-codexcodexlanguagemodelcodegenprogramsynthesis
salesforce/progen
Jun 2022 - Jul 2022
Official release of the ProGen models
Role in this project:ML Engineer
ML EngineerContributions:62 commits, 1 PR, 25 pushes in 1 month
Contributions summary:Erik's commits primarily focus on modifications to the `sample.py` and `likelihood.py` files, suggesting a focus on model evaluation and inference within the ProGen model. The user implemented sampling functionalities and explored log-likelihood calculations, indicating involvement in model usage and performance analysis. They also updated the model configuration file for ProGen2.
language-modelproteingenerative-model
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.