Machine learning engineer building agentic systems and digital twins. Currently shipping SimPilot, a multi-agent platform for engineering automation.
I work on agentic AI, the machine learning running underneath it, and the softwarethat holds the two together. PhD in computational modeling, mostly on reinforcement learning, generative models, and neural surrogates. Day job is Senior R&D at Synopsys, building neural-ODE digital twins. Off hours, I'm building an autonomous engineering agent.
- 2025—01
SimPilot
Multi-agent platform for engineering automation
A multi-agent system that prepares, runs, and analyzes engineering simulations autonomously. 155+ typed tools, deep-research loop, sandboxed execution. Full-stack TypeScript solo build.
- 202402
JAX evaluation harness
1000× faster RL evaluation
GPU-parallel rollout and metrics framework for evaluating RL policies. Cut a week-long sweep to under a day, which changed which experiments I was actually willing to run.
- 202303
Diffusion digital-twin surrogates
Conditional diffusion as a fast PDE solver
Generative digital twins for Navier–Stokes and Gray–Scott systems. >100× speedup over numerical solvers at comparable accuracy, with an evaluation framework I built alongside the model.
- Ongoing04
Scientific ML Notes
Open notebook
Working notes on scientific ML: PINNs, neural ODEs, operator learning, RL for control. Written for the version of me from three years ago.
- 2025—Senior R&D Engineer — ML & Digital Twins· Synopsys
Neural-ODE digital twins, ML evaluation infrastructure.
- 2021—25PhD, Scientific ML & Reinforcement Learning· University of Pittsburgh
RL agents, transformers, GPU-parallel evaluation in JAX.
- 2016—19Research Assistant — Hybrid ML Surrogates· VFE Research Institute
Neural-network surrogates for reduced-order modeling.
Things I keep coming back to: how to evaluate agents honestly (most LLM benchmarks are vibes), the gap between research papers and shipped systems, why scientific software is so much worse than it needs to be, and Persian poetry.
Currently reading The Information by James Gleick. Re-reading anything by Sutton when I want to think about RL clearly. Coffee: espresso, no sugar.
- Emailmhnaderi@live.com
- GitHubgithub.com/mhnaderi
- LinkedInin/mohammad-hossein-naderi
- CV/cv.pdf