Devanshi

Computer-vision assistive systems for low-vision users under Professor Kaveri Thakoor, plus ophthalmic support tools helping clinicians improve glaucoma and eye cancer diagnosis accuracy.

Built factor-based trading data pipelines processing 2TB of input daily, cutting insight turnaround time by 30%, and collaborated on an algorithmic trading platform integrating strategy optimization and multi-market execution.

Mathematical solutions for converting raster animations to the Lottie format, using Bezier-curve representations with OpenCV to interpolate additional frames per second.

Engineered an Azure-based generative AI meeting-analysis tool that cut manual effort for 3 clients by about 30 minutes per meeting; diagnosed defects and tracked issues through JIRA.

Evaluated 20+ companies across tiers and operating models, modeled growth trajectories with linear regressions, and contributed to a 15% rise in client engagement.

Pitched integrated services like accelerometer triggers with roadside assistance, reducing vehicular accident probability by 2.5%.

Monte Carlo stochastic simulations for estimating distribution properties, presented as a research paper at conference.

Investigated 23 space technologies and applications alongside senior scientists, with research presented on the CERN website.

Selected to support a 40-student cohort under Professor Gary Zamchick, hosting office hours and teaching prototyping methods for user-centered solutions.

Built a sentiment analysis and data extraction tool on financial data using BERT and GPT-3. Shut down for lack of traction, which taught me more than most things that worked.