Wrote feature engineering and evaluation pipeline using tools such as NumPy, Pandas, and RdKit

Part of Harvey Mudd CS/Math Clinic senior capstone.

Trained image classification and object-detection neural networks with Fastai

Deployed PyTorch models as inference endpoints using AWS Sagemaker

Processed RTSP and USB video streams using Deepstream SDK built on Gstreamer in Python

Wrote Python functions to extract attributes (eg: image brightness) from training and validation data

Implemented error pipeline in our cloud ML pipeline to correlate data attributes with accuracy metrics like loss or mAP in order to create better training and validation sets.