Want to know if this job is worth applying to?
Mountain View, California, United States
Onsite
About Unconventional
We're a team of researchers and engineers, working across AI theory, silicon architecture, system modeling, agentic software, and AI systems. We tape out test chips every four months and release new models every six months. If that cadence sounds intense, it is. We're taking a big swing with big risk, and we need people who want to go deep on hard problems.
The Role
We're looking for a Technical Program Manager to own hardware-software co-development across our test chip program. You'll be the connective tissue between silicon, architecture, systems, and software teams—driving rapid prototyping cycles that let us validate new AI architectures quickly and learn fast.
This role leads to owning program execution for our first product.
Responsibilities
Drive test chip programs from architecture exploration through tapeout, bring-up, and characterization. You'll manage aggressive 4-month tapeout cycles, coordinating across silicon, hardware architecture, and systems teams.
Own hardware-software co-development schedules and dependencies. Our architecture is novel, which means software, compilers, and system integration are deeply intertwined with silicon decisions. You'll keep these workstreams synchronized.
Build and maintain the program infrastructure for rapid iteration—tracking experiments, capturing learnings, managing risk, and making tradeoffs visible so the team can move fast without losing coherence.
Work across AI theory, system modeling, and silicon teams to translate research directions into executable program plans. You'll need to understand what we're trying to learn from each test chip and structure the program around those goals.
Prepare the organization for product execution. As we move from prototyping toward our first product, you'll help evolve our program discipline from research velocity to production readiness.
What We're Looking For
Minimum qualifications
Preferred qualifications
Why Join Us?
We're not building a better Von Neumann computer. We're building a different kind of computer based on a different understanding of what AI computation requires. The TPM role here isn't about tracking Gantt charts—it's about enabling a team of researchers and engineers to iterate fast on ideas that don't exist anywhere else.
