Nir Hutnik
NLP LLM Research Team Lead at DeepKeep
Israel
Summary
🤩
Rockstar🎓
Top SchoolNir Hutnik is an NLP and LLM research leader with 4 years of focused industry experience building and evaluating large language models and ML validation tools. Currently leading the NLP/LLM research team at DeepKeep, he develops real-time protection algorithms for LLMs, automated evaluation suites, and fine-tuning strategies like LoRA for tasks such as PII and DoS detection. Previously at Deepchecks he led open-source efforts to create rigorous tests for ML data and models—including contributions to a popular validation repo where he implemented feature-contribution checks using Predictive Power Scores. His background in applied mathematics and operations research from service in the IDF gives him a quantitative edge for designing robust detection of drift, shortcuts and hallucinations. Known for a love of puzzles and mathematical thinking, he blends rigorous research with practical product-driven engineering to move models safely into production. Based in Israel, he combines hands-on coding and open-source impact with team leadership across research and engineering.
4 years of coding experience
5 years of employment as a software developer
Bachelors, Applied Mathematics, Bachelors, Applied Mathematics at Bar-Ilan University
Stackoverflow
56reputation
1kreached
2answers
0questions
Github Skills (13)
machine-learning10
model-validation10
python10
data-science10
pytorch9
computer-vision9
keras6
visualization6
validation6
random-forest6
cross-validation6
scikit-learn6
resampling6
Programming languages (2)
Github contributions (4)
deepchecks/deepchecks
Oct 2021 - Jan 2023
Deepchecks: Tests for Continuous Validation of ML Models & Data. Deepchecks is a holistic open-source solution for all of your AI & ML validation needs, enabling to thoroughly test your data and models from research to production.
Role in this project:
Data Scientist & ML Engineer
Data Scientist & ML EngineerContributions:565 reviews, 132 commits, 182 PRs in 1 year 3 months
Contributions summary:Nir's primary contribution involved implementing the `single_feature_contribution` check for the `deepchecks/deepchecks` repository, a tool designed for validating ML models. This includes the creation of the `SingleFeatureContribution` class and associated plotting utilities, which leverages the PPS (Predictive Power Score) to assess a feature's ability to predict the target variable. The user also developed related tests to ensure functionality. The changes include the creation of a new file, addition of metrics, and adjustments based on new library features.
ml-modelspythonjupyter-notebookdata-modelsmodel-monitoring
deepchecks/monitoring
Jun 2023 - Jun 2023
Deepchecks Monitoring- Continuous Validation of ML Models & Data In Production.
Contributions:2 reviews, 1 comment in 1 day
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.