Machine Learning Engineer Intern
At Rockfish, I lead initiatives to streamline ML onboarding by benchmarking workflows, ranging from manual setups to automated pipelines with hyperparameter sweeps, data validation, and model recommendations, that are tailored to diverse use cases. I built automated dataset classifiers (tabular vs. time series) that dynamically inform the ingestion engine for optimal model selection. For training, I implemented sample-based hyperparameter tuning for GANs and Transformers, significantly reducing compute costs while maintaining performance. I also optimized epoch scheduling to improve rare category representation in synthetic data. To ensure data fidelity, I developed robust evaluation metrics and improved Rockfish’s synthetic data quality by 12% over top competitors like Gretel and Mostly AI.
