At Handshake AI, I trained and fine-tuned large language models in the science domain, creating prompts and reference responses and evaluating outputs for factual correctness, reasoning, and completeness. I also identified recurring errors and edge cases and provided structured feedback for iterative fine-tuning.
In my projects, I built machine learning workflows for customer churn and used-car price prediction, and analyzed NYC taxi trips with SQL. I also developed a CNN to classify handwritten digits using the MNIST dataset.
