Fangyu Wu

Hello!
This is Fangyu Wu.

I am a postdoctoral fellow at Columbia working on AI, operations, and robotics. I combine foundation models, control, and optimization to develop autonomous systems, including automated vehicles, robotic manipulators, and AI agents. I received my Ph.D. in EECS from UC Berkeley and previously interned at NVIDIA.

I am on the job market this fall.

You can find my latest CV here. I am best reached at You can also find me on Google Scholar and LinkedIn.

Invited Talks

  1. INFORMS · Applied Probability Society

    Efficient Optimization of Autonomous Agents with Differentiable Language Models

    MA50 - Recent advances in data-driven decision-making Scheduled · · Session 8:00–9:15 AM PT · Moscone South-313 (Level 3)
  2. Columbia · DRO

    Can language models dream in continuous think?

  3. Columbia · CEEM

    Control of Multi-Agent Dynamical Systems with Applications to Mixed-Autonomy Transportation

  4. Amazon · New York

    Learning to Accelerate: A Method for Speeding Up Costly Computation

  5. HKU · DASE

    Control of Multi-Agent Dynamical Systems with Applications to Mixed-Autonomy Transportation

Show all 18 talks Show fewer talks
  1. Cornell · CEE

    Introduction to Connected and Automated Vehicles and Field Experiments

  2. Columbia · DRO

    Control of Multi-Agent Dynamical Systems in Understructured and Hyperstructured Environments

  3. University of Washington · CEE

    Control of Multi-Agent Dynamical Systems with Applications to Mixed-Autonomy Transportation

  4. Cornell · Info. and Decision Sci. Lab

    From Distributed Control to Centralized Optimization: Charting the Next Decades of Autonomous Driving

  5. Caltech · AMBER Lab

    From Distributed Control to Centralized Optimization: Charting the Next Decades of Autonomous Driving

  6. NVIDIA · Autonomous Driving

    An MPC Approach to Longitudinal Trajectory Planning

  7. ICRA · Workshop on Perception, Action, Learning

    Motion Planning in Understructured Road Environments with Stacked Reservation Grids

  8. Berkeley DeepDrive Virtual Workshop

    Navigating through Chaos: Planning with Stacked Reservation Grid

  9. CPS PI Meeting

    An Open Robotics Platform for Multiagent Learning and Control

  10. ITSC · Conference Presentation

    Connections between Car-Following Models and Neural Networks

  11. ITSC · Deep RL and Transportation Tutorial

    A Tutorial on Simulation of Urban Mobility

  12. DDETFP Research Showcase

    Hybrid Microscopic Traffic Modeling Using a Classic Car Following Model and a Corrective Neural Noise Model

  13. TRB Annual Meeting

    Fuel Consumption in Oscillatory Traffic: Experimental Results

Research Areas

My research spans Physical & Agentic AI, Control & Optimization, and Automation & Mechatronics. Recently, I focus on differentiable latent-space models, memory for VLA models, and LLM-assisted algorithm synthesis for autonomous systems. See the full publication list.

Here are a few recent examples of my research in practice:

  • I contributed to Agents’ Last Exam, a benchmark used to evaluate models like GPT-6 Astra and Claude Opus 5.
  • I helped KNAPP, a global warehouse automation leader, achieve 58.9% higher throughput than the production baseline.
  • I contributed to adaptive cruise control in NVIDIA DRIVE AV, used in Mercedes-Benz production vehicles.