Researcher profile
Mehrdad Zaker, Ph.D.
Founder, Neural Intelligence Labs · San Francisco + Chicago
I study how agents learn to act in consequential software environments—and how planning, state, rewards, and verification shape what they can reliably do.
Focus
Current work centers on reinforcement-learning environments for software engineering, terminal, and web tasks; reliable long-running agent systems; and planning under operational constraints.
Research path
Mehrdad’s work connects planning, reinforcement learning, human–robot collaboration, and explainable intelligent systems with deployable agent infrastructure. For his full biography, career history, and writing, visit his personal site.
Now
- Executable RL environments for SWE, terminal, and web agents
- Reward contracts and verifier robustness
- Operational memory, state, evidence, and recovery
- Open research collaboration in San Francisco + Chicago