Vyom U.

• Designd a pipeline to extract key features from a video to predict if surgery would be successful. • Publication - Vyom Unadkat, Dhiraj J Pangal, Guillaume Kugener, et al. Machine Learning Analysis of Intraoperative Video Without the Use of Code, 2021 Congress of Neurological Surgeons Annual Meeting. (Submitted).

• Automated manual process of obtaining data from multiple vendors by establishing channels improving efficiency by 25%. • Created a cloud based Datawarehouse on Snowflake for Data Analysis.

USC Melady Lab • Worked as a Researcher and developed iOS app integrated ML model to anticipate surgical candidacy based on burn images. • Publication - S. Huang, S. Rambhatla, L. Trinh, M. Zhang, B. Long, M. Dong, V. Unadkat, H. Yenikomshian et al. DL4Burn: Burn Surgical Candidacy Prediction using Multimodal Deep Learning, AMIA 2021 Annual Symposium. (Submitted).

• Solved Spaghetti problem in 3D printing by developing a runtime surveillance and analysis module deployed on Raspberry Pi. • Reduced material wastage and production time by 25% by distinguishing discrepancies while product is being printed.

• Optimized process of video and audio alignment for multiple languages based on clips to be censored for Viacom US. • Incorporated Video Encoding algorithms for generating feature vectors and detecting clips to be deleted or censored. • Delivered an algorithm saving 100+ man hours daily requiring only human supervision.

• Designed an iOS application called MySA by integrating ML leveraging REST APIs increasing user engagement by 20%. • Crawled online activities to retrieve data, analyze and perform sentiment analysis to assign a score to each user.