Sept. 2025 – Present

• –Improved the accuracy of comic-book translation for four south-Asian dialects by 25% with image-based Retrieval-Augmented Generation.
• –Decreased expected translation time by 50% by reducing human intervention on incorrect translations.

Jun. 2025 – Aug. 2025

• –Researched the use of custom semantic linear classifiers for improving zero-shot model prediction.
• –Modified OpenAI's SOTA Contrastive Language-Image Pretrained (CLIP) model, improving mAP on the human action recognition dataset from 60% to 93%.
• –Decreased finetuning time by 3x with PyTorch DDP (Distributed Data Parallel) and integrated Weights & Biases (W&B) for efficient experiment tracking and debugging.

Jan. 2023 – Aug. 2025

• –Conducted research characterizing the macaque placenta and fetal brain, serving as references for studying inflammatory and infection dynamics during the third trimester.
• –Developed pipeline that automates preprocessing and downstream analysis of single-cell RNA datasets, reducing cell annotation processing time 90+% (7+ days to <12 hours).
• –Leveraged statistical analysis and research findings to help successfully secure $200K in grant funding.

Jan. 2025 – Present

• –Created a fully automated outreach pipeline using Python, Playwright, Google Maps API, and Gemini 2.5 Flash to scrape and personalize outreach for 700+ businesses, improving sponsor response success by 5x.
• –Launched columbiawushu.org with animations, embedded videos, and automated email systems, increasing web activity 4x and reducing administrative email traffic by 70%.