Yugraj
Some things about me
- I'm a Computer Science senior at UC Davis, graduating June 2026.
- Currently researching LLM safety — studying bias and jailbreak vulnerabilities — at the C² Computational Lab.
- Working on people analytics at Amazon, figuring out how to make operations strategy decisions with data.
- I like hard problems: predicting congressional votes, detecting adversarial attacks, building infrastructure that scales.
Some things I'm interested in
- Machine learning and adversarial security. I'm interested in building systems that are robust against attacks.
- Cloud infrastructure and serverless architecture. I like building things that scale without babysitting.
- Data-driven prediction systems. Using data to forecast outcomes in politics, gaming, and beyond.
- Side projects and shipping fast. Sometimes the best learning happens outside the main gig.
Experience
Software Engineer Intern · UC Davis Center for Animal Disease Modeling and Surveillance
- Developed NLP pipelines mapping 30+ years of vet records to Vet-ICD-O codes for cancer registry analysis.
- Designed a hybrid annotation pipeline (regex + LLM cascade) to generate training labels from unstructured veterinary diagnoses, replacing manual labeling across 43K records.
- Co-authored peer-reviewed abstract accepted at SAVY 3.0 (Cornell) on veterinary cancer classification via NLP.
Co-Founder · DavisDineBuddy
- Shipped a dining app used by 50+ daily active students, with automated menu scraping across 4 campus locations.
- Released iOS and Android clients simultaneously on a shared GCP/Firestore backend, reducing duplicate engineering effort across platforms.
- Developed a Claude-powered meal planner that reduced menu browsing by generating weekly plans from user preferences, meal history, and macro goals.
Software Engineer Intern · American Association of University Women, Davis
- Led a full website redesign as sole developer, increasing monthly traffic by 300% through a modernized UI.
- Served as IT contact for the branch, managing website, member communications, and technical infrastructure.
Undergraduate Research Assistant · C² Computational Lab, UC Davis
- Automated relative risk scoring across 50+ topics with Pandas, cutting analysis time from hours to seconds.
- Created Matplotlib visualizations analyzing whether platform-owned LLMs favored their parent platforms, guiding research on model transparency.
Projects
- QUORUM — congressional vote prediction system analyzing 75+ features per legislator, with 95.7% accuracy on partisan votes.
- VANTAGE — real-time Valorant analytics platform on a serverless stack, tracking player performance and agent meta.
- DARWN — adversarial cloud security simulation running automated red team scenarios against AWS infrastructure.
- Pyrosphere — wildfire detection combining YOLOv8 vision with real-time weather data for early warning alerts.