Data science · Machine learning · AI
Idan Horowitz
I’m a data scientist and machine learning practitioner with a passion for AI. I own projects end to end: framing the problem, taming messy data, building and deploying the model, and evaluating it once it is live. I thrive on finding novel solutions and on understanding how and why they fail. Have a hard problem that needs taking from idea to production? Get in touch.
Selected work
-
01
Neighborhood Smoothing for Calibration
Developed NeSCal, a novel PyTorch training objective that improves deep learning classifier calibration by smoothing predictions across learned neighborhoods.
-
02
CLARA | Multi-Agent Loan Risk Assessment Prototype
Loan-assessment prototype combining predictive models, retrieval, LLM agents, and evaluation-driven revisions.
-
03
Do midseason manager replacements improve football results?
An observational causal-inference study of Premier League managerial changes.
-
04
The Future of Typing
A concept keyboard for augmented reality that replaces the QWERTY layout with LLM word and phrase suggestions, hands-free pointing and speech input.