- Streamlined climate project evaluation by developing a decision-support tool with PyTorch, Transformers, and Dash, processing 13,000 project entries.- Achieved an F1 score of 0.84 with a predictive model trained on diverse project-level data and external indicators, effectively assessing the potential impact of hypothetical international climate development projects.- Developed an NLP pipeline for text-network and sentiment analysis, enhancing insights from project reports.

- Engineered 3 trading algorithms, adopted by 7 investors on a leading Japanese platform, managing over $25,000.- Yielded a 160% return rate over 3 years, showcasing knowledge adaptability and algorithm development skills.

- Developed an image processing, text extraction and fuzzy-matching pipeline, digitizing and labeling 200,000+ archival documents,- Implemented a data validation tool, cutting down mislabeled facial recognition instances by 25%, thereby improving dataset quality and model accuracy.- Provided software and hardware support to internal and external users.