Shachar Heyman
Data Scientist & ML Engineer
Berlin, Germany
About
Data Scientist & Machine Learning Engineer with 8+ years of experience building production AI systems across computer vision and LLM-based platforms. Experienced in designing end-to-end ML pipelines, evaluation frameworks and adaptive learning systems that translate research into scalable products. Proven track record of leading ML initiatives from early-stage development to production deployment in both startups and enterprise environments.
Experience
Convergent-AI
Data Science & AI Consultant (Contract)
2025 - Present
Designed the ML architecture for an enterprise training platform based on LLM-driven human-interaction simulations. Built evaluation pipelines to assess conversational performance and developed benchmarking methodologies for measuring human-likeness of AI-driven conversations.
Siemens
Machine Learning Engineer / Data Scientist
2023 - 2025
Designed and deployed deep learning systems for industrial visual inspection, improving defect detection accuracy in production environments. Built end-to-end ML pipelines from data ingestion to model deployment and led generative data augmentation initiatives.
Inspekto (acquired by Siemens)
Machine Learning Engineer / Data Scientist
2017 - 2023
Built end-to-end ML pipelines from data acquisition to deployment and monitoring. Developed computer vision models for automated defect detection using CNNs. Implemented production-grade CI/CD pipelines for reliable model delivery.
Tel Aviv University
Teaching Assistant
2018 - 2021
Supported undergraduate mathematics courses including Linear Algebra, Set Theory, Combinatorics & Topology.
Education
Tel Aviv University
B.Sc. Mathematics & Philosophy
2017 - 2020
Stanford University
CS231n: Deep Learning for Computer Vision
2021
Goethe Institut & Sprachsalon Berlin
German Language (A1-B2)
2024 - 2026
Projects
Auto Document Writer
Aug 2025 - Present
SQL-based product for Time Prints (film production company), auto-filling filming-day-reports from MySQL database, saving hundreds of work hours annually.
SQL, NLP
Shift Scheduler
GitHubOct 2025
Shift assignment tool for a documentary film festival. Normalized free text using offline LLM, processed with Pulp for optimal scheduling across multiple locations.
Python, Pulp, LLM
Nena Song Generator
GitHubJul 2025 - Sep 2025
Auto-regressive generative model that creates Nena-style song lyrics from a given title. Trained on scraped lyrics.
NLP, LLM, Python
Scrape, Archive & Classify
Jul 2025
Automated tool for downloading images, extracting numbers via OCR, renaming files, and generating annotated spreadsheets. One-command solution for repetitive tasks.
Data Scraping, OCR, Python