Hi, I'm Ali!

I'm a PhD student at Michigan State University. I developed Optimal Inferential Control (OIC) for scalable and efficient control of deep learning dynamical systems. I obtained my M.Sc. at the University of Kansas, where I developed heavy-tailed Bayesian inferential motion planning (BIMP) for robust decision-making under uncertainty.

I have also completed internships at Cruise/General Motors and Mitsubishi Electric Research Laboratories (MERL).

Industry experience

Cruise/General Motors — AI/ML SWE Intern, 2026: Large-scale ML for autonomous-driving simulation; built large-scale data and distributed multi-GPU training pipelines for highway vehicle motion reconstruction. Large-scale ML for autonomous-driving simulation: developed large-scale data and distributed multi-GPU training pipelines for highway vehicle motion reconstruction.
Mitsubishi Electric Research Laboratories — Research Intern, 2024: Physics-based ML for thermofluid systems; Neural surrogate models that are physically consistent and scalable.

Selected Projects Get in touch

Selected projects

All projects

Cruise — General Motors · 2026

Vehicle fleet motion reconstruction using ML

Scalable scene reconstruction for recorded-driving simulations. A large-scale data pipeline processed terabytes (TBs) of tracking data across 38,000+ highway scenes, paired with a distributed training pipeline.

Industry experience

24%

Lower position error

Best-candidate average position error vs. the team’s prior baseline on approximately 58,000 held-out vehicle trajectories.

Large-scale training

12 → 3 hours

Before

12 hours

After

3 hours

Multiple H100 graphics processing units (GPUs), PyTorch DistributedDataParallel (DDP), the NVIDIA Collective Communications Library (NCCL), and mixed precision.

Simulation of a soft-robot manipulator following a reference trajectory

Continuous-time control

Control of ML-based dynamics

Continuous-time inferential control of neural ordinary differential equation (Neural ODE) models, including a soft-robot manipulator.

Up to 200× faster than the Interior Point Optimizer (IPOPT) on GPU in tested control benchmarks.

Control results

Simulation of vehicle overtaking with Bayesian motion planning

Autonomous vehicles · Motion planning

Planning under uncertainty

ML-based vehicle dynamics modeling and Student’s-t distributions for motion planning with non-Gaussian uncertainty and outliers.

ML-based vehicle dynamics models trained on real driving data.

Planning results

Vapor-compression HVAC system used for physics-constrained modeling

Mitsubishi Electric Research Laboratories · 2024

Learning thermofluid dynamics from limited data

A physics-constrained model combining Neural ODEs and gated recurrent units (GRUs) for vapor-compression heating, ventilation, and air conditioning (HVAC) systems.

14%Better forecasting accuracy

5.7×Runtime speedup

Compared with the prior modeling approach.

Read the paper

Research