Persian Ink
An offline Persian handwriting keyboard for iPad. Write Farsi with Apple Pencil, review recognition candidates, and insert text into Notes or another supported app.
Independent AI research · San Francisco + Chicago
We build and study reinforcement learning environments for software, terminal, and web agents. Our research focuses on resettable tasks, observable state, trustworthy verification, and recovery from failure.
Explore NeuralInt apps for Persian writing, focused work, family routines, and fashion discovery, alongside open-source tools.
An offline Persian handwriting keyboard for iPad. Write Farsi with Apple Pencil, review recognition candidates, and insert text into Notes or another supported app.
An offline Persian and Farsi swipe keyboard for iPhone and iPad, with tap typing, نیمفاصله, word suggestions, and learning from your corrections. No Full Access required.
A focus timer for iPhone that turns completed work sessions into a simulated national-park hike, with optional Screen Time distraction blocking, soundscapes, and Lock Screen Live Activities.
A family routine and reminder app for parents. Write your own announcements, schedule iPhone reminders, and tap to play speech on a compatible Google, Nest, or Cast-enabled speaker.
An AI fashion discovery app for iPhone. Search for clothing with a description or reference photo, explore products across stores, and preview styles with virtual try-on.
An alpha workspace for coordinating coding agents with isolated Git worktrees, reviewed integration, and saved sessions.
RadioDesk is an open-source workspace for amateur-radio station control, signal analysis, RF tools, and local logging.
Better agents require more than better policies. They need worlds with legible state, meaningful consequences, trustworthy feedback, and enough variation to learn skills that transfer.
Resettable, observable worlds for training and evaluating agents across repositories, terminals, and browser-based workflows.
Testable outcome contracts, partial-credit signals, and adversarial checks that distinguish task completion from reward hacking.
Task generation and curricula that transfer skills from shell operations to software engineering and stateful web work.

Founder & researcher
Reinforcement learning · Planning · Agent systems
Mehrdad is a computer scientist and founder of Neural Intelligence Labs. His work connects reinforcement learning, planning, and human-centered agent systems with the systems problems that make long-running AI work reliable in practice.
Open to research collaborations, benchmark partnerships, visiting talks, and working sessions in San Francisco + Chicago around agent environments and reliable AI systems.
Start a conversation