Yash Patel
Scholar

Yash Patel

Google Scholar ID: _BQwMtgAAAAJ
PhD Student, University of Michigan
robust decision makingoptimizationconformal predictionneural operators
Citations & Impact
All-time
Citations
67
 
H-index
5
 
i10-index
2
 
Publications
14
 
Co-authors
0
 
Resume
Academic Achievements
  • PhD research focused on principled uncertainty quantification, robust decision-making, and AI for Science
  • Undergraduate projects include: HyperLoop Pod design (Top 30 in SpaceX competition twice), senior thesis on deanonymizing Bitcoin via graph clustering, Tesla Autopilot analysis, Neural Branch Predictor
  • At Meta: developed real-time dynamic object reconstruction (72 FPS), scene reconstruction on HMDs, and optimized PyTorch models for Qualcomm SoCs
  • At Bose: novel transformer-based Wiener filter adaptation surpassed FxLMS baseline
  • At Waymo: rare-event sampling algorithm dramatically accelerated simulation pipeline
Research Experience
  • Jan 2026 – Present: Research Engineer, Harmonic
  • Nov 2025 – Present: Research Engineer (AI Safety), Anthropic — working on alignment and reliability of frontier models
  • Jun 2025 – Sep 2025: Data Science PhD Intern (Simulations), Waymo, advisor Aman Sinha — implemented ADMM-based distributed convex optimization for rare-event importance sampling, achieving 20x speedup
  • Jan 2025 – Jun 2025: Machine Learning Research Co-op, Bose, advisors Russell Izadi and Shuo Zhang — implemented SAC/PPO for adaptive-FIR noise cancellation; developed novel transformer-based Wiener filter adaptation outperforming FxLMS by 10% dB reduction
  • Jul 2018 – Sep 2021: Senior Software Engineer (IC5), Meta (Facebook), advisor Albert Parra Pozo — worked on real-time 3D reconstruction, rendering, and on-device ML optimization
  • May 2015 – Aug 2015: Research Intern, Columbia University