Practical experience in data quality. A drive to advance reliable AI.
Third-year Industrial Engineering & Information Systems student. Currently a Market Researcher at Lusha.
Hands-on experience in data quality, model evaluation, and research-driven process design. Ready to help the lab investigate model failures, build reliable evaluation methods, and turn research questions into practical experiments.
Evaluated 30+ CV models and led 12 annotators. Traced data and workflow failures with Data Science, contributing to a 23% precision improvement on a model that had challenged the department for three years.
Making data more useful through systematic research
Researched five technology companies and academic sources, then built five processes that doubled work speed and cut required annotations by 67%.
Turning open questions into testable workflows
Redesigned CV model training and management workflows, increasing output by 50% and reducing recurring manual work by 80%.
In development
02 | Tag · AI advisor for data labeling and model evaluation
Making expert knowledge useful to an AI agent
Tag is an AI advisor for data labeling, quality, and model evaluation that I am co-developing with Danielle Menuhin, founder of Data Operations IL. We are building and validating its knowledge base with industry leaders to provide source-grounded answers, surface assumptions, and acknowledge knowledge gaps.
Early version, shared for your review. Please don’t forward the demo. Feedback is very welcome.
03 | Research Interests
What interests me
Developing accurate and reliable methods for evaluating AI models
What drives me
Industry conversations and meetups have exposed me to subjective evaluation methods, motivating me to develop rigorous, reliable approaches that provide clear, well-founded answers.
What interests me
Making data more accessible and useful for models and agents
What drives me
Building Tag and working with AI tools has shown me how strongly wording and information structure affect model outputs. Even asking a model to reason before categorizing can improve how well it uses the same data.
04 | References
“What most impressed me about Shaked is his fast learning skills and his capability to report to managers from three different departments while delivering outstanding work.”
Moti CohenSr. Project Manager
“With remarkable attention to detail, Shaked contributed a significant part to our successful new product launch. Thanks to his rare communication skills, our entire organization was much more motivated.”