I build algorithms that control real-world systems, pairing the speed of machine learning with the guarantees of control theory. Specifically, I design neural operators with provable guarantees for systems ranging from PDEs and robots to biological systems. I'm thankful to be advised by Yuanyuan Shi and Miroslav Krstić, and supported by the DOE Computational Science Graduate Fellowship.
The path (so far...)
-
2018 – 2022Vanderbilt University
I did an accelerated B.S./M.S. majoring in physics, applied mathematics, and computer science. At Vandy, I made my first exploration into research, working on laser-driven electron dynamics, and took my first steps in reinforcement learning and robotics. I also learned to code the old-fashioned way (without Claude or GPT), interning at both T-Mobile and MongoDB. I will forever cherish my Vanderbilt years!
-
2022 – nowUC San Diego
Ph.D. in Electrical & Computer Engineering with Yuanyuan Shi and Miroslav Krstić, developing neural operators with control-theoretic guarantees. DOE Computational Science Graduate Fellow.
-
2024Lawrence Berkeley National Laboratory
Randomized linear-algebra solvers for distributed nonconvex optimization.
-
2025Amazon
Custom LLMs adjusting time-series demand forecasts across 30M+ products.
-
2026Microsoft
Research scientist intern building foundation models for the power grid.
Education
Honors
- Best Paper Finalist, Learning for Dynamics & Control (L4DC) 2025 (3 of 119)
- Commitment to Diversity Award, UC San Diego ECE
- Underwood Memorial Award, Vanderbilt Physics & Astronomy (outstanding graduating senior)
- Best Undergraduate Paper, Vanderbilt Physics & Astronomy
Grants & fellowships
- DOE Computational Science Graduate Fellowship (four-year PhD support, $400k+)
- NVIDIA Academic Grant, co-investigator (8 RTX Blackwell GPUs donated to support PDEControlGym, ≈$150k)
- First-Year Ph.D. Fellowship, UC San Diego ECE (≈$100k)
Industry & national-lab experience
Leadership & service
- Organizer, PDE-AI Workshop on AI for Modeling, Control, and Optimization of PDEs, ACC 2026
- Co-Instructor, Physics-Informed Machine Learning, UC San Diego (see my teaching)
- Reviewer for Automatica, IEEE Transactions on Automatic Control, Systems & Control Letters, ICLR, L4DC, CDC, ACC, and AAAI
Mentoring
I mentor hands-on, meeting each student in weekly one-on-ones and working alongside them from framing the problem to the final result.
-
Peter Quawas
Built the robotic-manipulator simulations showing the neural-operator predictor holds accurate tracking even when the state is measured only at sparse, irregular, and noisy times.see publication →
-
Filip Bajraktari
Led this paper under my mentorship, developing both the theory and the experiments for compensating several distinct input delays at once, demonstrated on a car-like vehicle with different delays on its steering and acceleration.see publication →
-
Peijia Qin
Developed the manipulator experiments for our Best Paper Finalist work, where a learned predictor tracks a target signal under a long actuation delay that would otherwise destabilize the loop.see publication →
-
Gabriel Bortoni
Implemented the neuron-model (Stefan) environment in PDE Control Gym, our open benchmark for data-driven boundary control.see the project →
-
Gokul Gandhikumar
Added the ARZ traffic-flow environment to PDE Control Gym, extending the benchmark to congestion-control problems.see the project →
-
Evan Wu
Built the glioblastoma tumor-growth environment for PDE Control Gym, bringing a medical boundary-control problem into the benchmark.see the project →
-
Yunhao Li
Investigated the dynamics of neural ODEs and delivered a literature review on learning-based control to our research group.
Outside research
Outside the lab, most of my time goes to my dog, Erwin, and, lately, to learning ice hockey as an adult, despite barely being able to skate when I started.