Physical Intelligence
1 open Forward Deployed Engineer role at Physical Intelligence. Curated with comp bands.
On the board since Jul 2026 · still hiring
Our take
Why Physical Intelligence
The rare FDE that's hands-on-hardware — first responder for a live robot fleet at one of the hottest robotics foundation-model labs, not embedding at a customer.
- FDE type
- Product-embedded
- Travel
- Low
- Comp
- Undisclosed
- Stage
- Frontier lab
A good fit if you're…
- Engineers who want to debug real robots in production, not customer codebases
- People drawn to vision-language-action foundation models at the frontier
- Builders comfortable being the first responder for a live robot fleet
- Candidates willing to work on-site in San Francisco, where the fleet physically lives
The GPT-for-robots lab
Physical Intelligence builds vision-language-action (VLA) models — control policies that take camera feeds plus a natural-language instruction and output real-time motor commands. Founded in 2024 by a team including CEO Karol Hausman, it licenses model software rather than building robots itself. Its π0 model, open-sourced in late 2024, controls seven robot types across 50+ tasks; the newer π*0.6 adds reinforcement learning from the robots' own experience and human corrections. Open weights made π0 one of the most-studied robot policies in the field, which is partly why PI can hire for fleet aptitude over big-robotics pedigree.
What “forward deployed” means here
Unusually, you deploy to the company's own robot fleet, not a customer's software. The Forward Deployed Robotics Engineer is the first responder when robots misbehave in production — triaging failures, instrumenting the fleet, and feeding real-world breakage back to the model teams. It's hands-on-hardware fleet operations, far closer to product-embedded reliability work than to the customer-facing enterprise deployment most FDE roles describe. In practice you live in logs, dashboards and the robots themselves — reproducing a failure, isolating whether it's perception, policy or actuation, and shipping the fix or the data that prevents it.
The role's unusual experience bar
The posting asks only about two years of experience and a B.S. — low for a frontier lab, and a deliberate signal rather than a low bar. Almost nobody has debugged multi-robot fleets in production, so PI is hiring for aptitude and hardware instincts over pedigree. The flip side: expect leveling and comp to be set case-by-case rather than slotted into a published ladder.
Comp: undisclosed, and genuinely so
Comp isn't posted on the board. As a heavily funded lab you can expect a meaningful equity component and competitive cash, but the modest experience ask means the band is wider and lower-anchored than the frontier-lab headline implies. Treat equity as the real lever, confirm the level you're being hired at, and negotiate specifics — very little here is standardized yet.
The capital behind the bet
PI raised a $600M Series B in November 2025 led by CapitalG at a $5.6B post-money valuation, more than double its $2.4B a year earlier; backers include Jeff Bezos, OpenAI, Thrive Capital, Lux and Bond. That's roughly $1B+ raised in under two years — ample runway, but also expectations priced for a category-defining outcome, which shapes how intense the operational pace is likely to be.
Who thrives, who bounces
Thrives: engineers energized by physical debugging, comfortable with on-site work and ambiguity, who want to sit close to both the model and the metal. Bounces: anyone optimizing for fully-remote, pure-software IC work or a defined promotion track — this role is early, operational, and robot-fleet-shaped, and that is not going to change soon.
Questions candidates ask
Do you deploy to customer sites?
No. Unlike a classic FDE, you deploy to Physical Intelligence's own robot fleet — the “customer” is internal. The role is on-site in San Francisco because the hardware is.
Why does it only ask for ~2 years of experience?
Because almost nobody has production multi-robot-fleet experience. PI is hiring for hardware-debugging aptitude rather than a long résumé, so expect leveling to be assessed individually.
Is this a software job or a robotics job?
Both, tilted toward hardware. You write code and instrument systems, but the day-to-day is diagnosing why physical robots fail in the real world and fixing it fast.
What's the compensation?
Not disclosed on the board. Expect competitive lab cash plus equity; because the experience bar is low, negotiate the band and the equity grant explicitly.
Analysis updated 2026-10-16
Open roles at Physical Intelligence
Compare Physical Intelligence
Others in the same tier
Free forever. Unsubscribe anytime.