
Robot labs & makers · Foundation-model lab
Physical Intelligence
About Physical Intelligence
Physical Intelligence (π) builds general-purpose robot foundation models. It hires non-engineer Robot Operators who teleoperate robotic arms through tasks to generate training demonstrations, measured explicitly on data-collection volume and quality.
Stated by the employer
, RobotTraining.jobs shows 1 Physical Intelligence job open.
Live Physical Intelligence roles
Updated 1 Nov 2026, 06:01 UTCRoles like these you can apply to today
Similar work at other employers.

Car-Mounted 360° Video Recorder (opens the job details)
micro1 · Demonstration Collection · Remote· 4w agoDetails
$15–$20 per accepted video-hour
At-Home Video Recorder (LATAM) (opens the job details)
micro1 · Egocentric Capture · Remote· 1w agoDetails
$6 per accepted video-hour
360° Video Recorder (Insta360 / GoPro MAX) (opens the job details)
micro1 · Demonstration Collection · Remote· 1mo agoDetails
$20 per accepted video-hour
At-Home Video Recorder (opens the job details)
micro1 · Egocentric Capture · Remote· 1w agoDetails
$15 per accepted video-hour
Joining through these links can earn us a referral fee. Your pay stays exactly the same.
Robot Operator (opens the job details)
Teleoperation · San Francisco, United States· 2mo agoDetails
$25/hr
More robot jobs in the cities where Physical Intelligence is hiring
Some links here are referral links: when you join an employer through one, RobotTraining.jobs may receive a commission. It costs you nothing, and the employer pays you exactly the same.
Physical Intelligence jobs by country
Its open roles that accept applicants in each country.
Get new Physical Intelligence roles by email
We read Physical Intelligence's job board every day and send you its new openings once a week, with comparable paid jobs at other employers. No cost, and you can stop any time.
Is Physical Intelligence a good place to work?
Researched October 2026 · 10 sourcesMostly. Physical Intelligence is a richly funded robot AI lab that hires operators as full-time employees with benefits, but it pays $25 an hour for quota-driven evening or overnight shifts, and no employee rating could be read. The pay is low for San Francisco: a flat $25 an hour, against MIT's $32.44 living wage for a single adult there, for 8-hour standing shifts from 4 PM to midnight or midnight to 8 AM, with data collection targets every shift. It pays $25/hr in checked postings (stated by the employer).
Actively listing 1 open role todayPhysical Intelligence (π) is a San Francisco robot AI lab founded in 2024 by former Google DeepMind researchers and Stanford and UC Berkeley academics, including CEO Karol Hausman, Sergey Levine, Chelsea Finn and Lachy Groom. Reports in November 2025 said it raised $600 million at a $5.6 billion value, led by Alphabet's CapitalG. TechCrunch counted about 80 staff in January 2026. In October 2026 it opened its models to other robot makers in a Pi API beta.
Robot Operators are full-time hourly employees (at least five days a week) at a flat $25 an hour, with health, dental and vision insurance, a 401(k), paid time off, paid sick time and commuter help. Shifts are 8 hours with a 30-minute lunch and two paid breaks. The post does not state pay frequency or any equity.
Metrics-based role with data-collection goals per shift; 8-hour standing shifts (stated in the posting, not a worker rating). (checked on jobs.ashbyhq.com, 28 Oct 2026)
1 open role on Physical Intelligence, independently checked 1 Nov 2026.
On-site in San Francisco, California; it has also posted technician roles in Fremont.
What people praise
- Benefits come with the hourly job: health, dental and vision insurance, a 401(k), paid time off, paid sick time and commuter help, on full-time employment rather than a contract.
- No degree or robot experience is required: the post looks for hands-on or lab work, dexterity and care for detail, and counts gaming with controllers as a plus.
- The money looks secure for now: TechCrunch reported in January 2026 that investors valued it at $5.6 billion and that it does not burn much cash, with most spending going to computing power.
What people complain about
- The operator post pays $25 an hour flat; MIT's living wage for a single adult with no children in San Francisco County was $32.44 when read in October 2026.
- It is a quota job: the post says you must meet set data collection goals each shift, the work is repetitive by design, and you stand at a workstation for the full 8 hours.
- Only evening (4 PM to midnight) and overnight (midnight to 8 AM) shifts are offered, on Mon to Fri, Wed to Sun or Sat to Wed patterns, so two of the three include weekend days.
- We could not read any employee review: Glassdoor and Indeed refused plain reads on October 29, 2026, and the one Blind thread about working there had no first-hand answer.
What has gone wrong
- November 2025. SiliconANGLE, citing The Information and Bloomberg, reported a $600 million round at a $5.6 billion value led by Alphabet's CapitalG, with Lux Capital, Thrive Capital, Jeff Bezos, Index Ventures and T. Rowe Price. The company had not disclosed the round and had announced no product.
- January 2026. Co-founder Lachy Groom told TechCrunch he gives investors no timeline for making money and hopes to grow headcount "as slowly as possible", and called hardware the hardest part: it breaks, arrives late, and safety concerns complicate everything.
- October 2026. On October 13 it opened a Pi API beta that lets other companies fine-tune and run its models on their own robots. On October 27 Bloomberg reported talks to raise about $1 billion at a value near $20 billion, led by Avenir and Lux Capital; the round had not closed.
Best for: People in or near San Francisco who like precise, repetitive hands-on work, can stand through 8-hour evening or overnight shifts and hit targets, and want a benefited full-time job inside a leading robot AI lab. Not for: Anyone who needs daytime hours, remote work or pay that covers San Francisco living costs on one income, dislikes being measured on output every shift, or cannot stand at a station for 8 hours.
Honest catches
- Pay here is stated by the employer, not measured live from a job feed, so treat it as a benchmark rather than a guaranteed rate.
Sources for this review
- Ashby API: Physical Intelligence's open postings with full text (44 roles), read Oct 29, 2026
- Physical Intelligence: Robot Operator, San Francisco ($25 an hour, shifts, benefits, metrics), read Oct 29, 2026
- Physical Intelligence: Shift Lead, San Francisco (supervises hourly, shift-based operators), read Oct 29, 2026
- Physical Intelligence: home page and research log (Pi API beta, Oct 13, 2026), read Oct 29, 2026
- SiliconANGLE: Jeff Bezos-backed Physical Intelligence raises $600M (Nov 20, 2025)
- TechCrunch: A peek inside Physical Intelligence, the startup building Silicon Valley's buzziest robot brains (Jan 30, 2026)
- Startup Fortune: Physical Intelligence is in talks to double its valuation to $20 billion, citing Bloomberg (Oct 27, 2026)
- Wikipedia: Physical Intelligence Inc. (founders, funding), read Oct 29, 2026
- MIT Living Wage Calculator: San Francisco County, California (updated Feb 15, 2026), read Oct 29, 2026
- Blind: "Physical Intelligence" interview thread (no first-hand answer), read Oct 29, 2026
What Physical Intelligence pays
Stated by the employer · updated 2026-10-28Researched rate card (stated by the employer), updated 2026-10-28
Source examples collected October 28, 2026. This envelope spans individual advertised offers with different duties, locations or shifts. It is not a company-wide typical wage. See current openings for updated terms. (checked on jobs.ashbyhq.com, 28 Oct 2026)
Only 1 of its 1 open roles list a rate, not enough to call any of them typical for Physical Intelligence, so the figure above is the researched one. Here is what each kind of work lists:
| Kind of work | Open roles | Rates stated | Highest rate |
|---|---|---|---|
| Teleoperation | 1 | $25/hr1 rate | $25/hr |
| Role | Work type | Researched pay |
|---|---|---|
| Robot Operator | Teleoperation | $25/hour |
Wondering where a number comes from? Read how researched rate cards are sourced and graded.
Kinds of work at Physical Intelligence
Open any of them for the day-to-day tasks, every employer hiring for it and market-wide pay.
First time doing this kind of work? Read the beginner guides or line up pay at every employer.
What our research records about Physical Intelligence
23 dated facts from public sources (the company itself, filings, regulators, courts and the press), each kept with the publication it comes from in one of our research reports. Newest first; the newest is from 27 Oct 2026.
- Robot model training data8
- In-house data collection6
- Industry funding5
- Simulation and synthetic data2
- Robot data deals1
- Robot data prices1
Physical Intelligence was in talks to raise around $1 billion, from investors including Avenir and Lux Capital, valued at a valuation near $20 billion (reported).
Source: Startup Fortune · In Industry funding
Physical Intelligence runs a Set Team that builds and maintains the physical rooms where its robots collect data, such as kitchens, living rooms and evaluation stations.
Source: Ashby (Physical Intelligence careers) · In In-house data collection
Physical Intelligence is hiring an operations recruiter for high-volume Robot Operator hiring, saying its operators, copilots, shift leads, technicians and props staff collect training data on-site in San Francisco and the teams are growing quickly.
Source: Ashby (Physical Intelligence careers) · In In-house data collection
Physical Intelligence Robot Operators either lead the robot with their own arms (the robot mirrors them) or use specialized controllers, and the job is metrics-based with data collection goals per shift.
Source: Ashby (Physical Intelligence careers) · In In-house data collection
Physical Intelligence advertises an on-site Robot Operator role in San Francisco that teleoperates robot arms to collect training data, paying $25 an hour on 8-hour evening or overnight shifts, minimum 5 days a week.
Source: Ashby (Physical Intelligence careers) · In In-house data collection
One hour of robot operation for data collection: $25/hour, paid by Physical Intelligence to Robot operators.
Source: Physical Intelligence (Ashby job board) · In Robot data prices
All 23 facts about Physical Intelligence
π0.7 is trained on data from many different robots, human data, and autonomous episodes collected by running various policies.
Source: Physical Intelligence · In Robot model training data
Physical Intelligence was in talks to raise about $1 billion, from investors including Founders Fund, Lightspeed Venture Partners and Thrive Capital, valued at a valuation exceeding $11 billion (reported).
Source: TechCrunch · In Industry funding
π0.5 (human-to-robot transfer): Physical Intelligence found that egocentric video of humans, recorded with wearable cameras, roughly doubled performance on tasks where robot data is limited once its models were trained on enough robot data.
Source: Physical Intelligence · In Robot model training data
Physical Intelligence raised $600 million, led by CapitalG, valued at $5.6 billion (reported).
Source: SiliconANGLE · In Industry funding
Physical Intelligence says robot foundation models today mainly learn from demonstrations people collect by hand, such as teleoperation; for π*0.6 it adds expert teleoperator corrections and the robot's own autonomous experience.
Source: Physical Intelligence · In Robot model training data
Scale AI said it is providing and enriching robot training data for Physical Intelligence through its Data Engine for Physical AI, which collects data from dedicated robots and human demonstrations at its San Francisco lab and through contributors worldwide.
Source: Scale AI · In Robot data deals
Physical Intelligence co-founder Sergey Levine argues on his blog that cheap stand-ins for real robot data, such as simulation or web video, work only in a few cases that match their designers' assumptions, likening them to a spork.
Source: Sergey Levine, Learning and Control (Substack) · In Simulation and synthetic data
Sergey Levine, a Physical Intelligence co-founder, writes that adding human and even simulation data to broad, representative real-world robot data is likely to help.
Source: Sergey Levine, Learning and Control (Substack) · In Simulation and synthetic data
For pi0.5, Physical Intelligence used about 400 hours of data from its mobile manipulators doing household tasks in about 100 different homes.
Source: arXiv (Physical Intelligence) · In In-house data collection
π0.5 uses about 400 hours of data from mobile manipulators doing household tasks in about 100 different homes, the slice of its training data closest to its test tasks.
Source: arXiv (Physical Intelligence) · In Robot model training data
97.6% of the training examples in π0.5's first training phase do not come from mobile manipulators doing household tasks but from other robots and web data.
Source: arXiv (Physical Intelligence) · In Robot model training data
Physical Intelligence released the code and weights of π0 as open source on February 4, 2025, saying 1 to 20 hours of data had been enough to fine-tune it to a variety of tasks.
Source: Physical Intelligence · In Robot model training data
Physical Intelligence raised $400 million, from investors including Jeff Bezos, OpenAI and Thrive Capital, valued at $2.4 billion post-money valuation.
Source: CNBC (syndicated by NBC New York) · In Industry funding
Physical Intelligence pre-trained its pi0 model on about 10,000 hours of dexterous manipulation data from its own robots (7 robot configurations, 68 tasks), on top of open datasets such as OXE, DROID and Bridge.
Source: arXiv (Physical Intelligence) · In In-house data collection
Physical Intelligence pre-trained π0 on about 10,000 hours of its own dexterous robot manipulation data, alongside open datasets such as Open X-Embodiment.
Source: arXiv (Physical Intelligence) · In Robot model training data
Physical Intelligence's own π0 pre-training data covers 7 robot configurations and 68 tasks (its launch blog describes the data as coming from 8 distinct robots); the mixture also includes the Open X-Embodiment data from 22 robots.
Source: arXiv (Physical Intelligence) · In Robot model training data
Physical Intelligence raised $70 million in its Seed round, led by Thrive Capital.
Source: Maginative · In Industry funding
How to get started on Physical Intelligence
Researched September 2026Apply on Physical Intelligence's Ashby job board for the on-site Robot Operator role in San Francisco, choose an evening or overnight shift and a five-day pattern, get through a recruiter screen and interview that tests manual dexterity, attention to detail and reliability, then start as a full-time hourly employee teleoperating robot arms to record demonstrations its AI learns from.
Who can join
- Able to work on-site in San Francisco, CA. You operate real robot arms in the lab, so there is no remote option.
- Full use of both arms and hands to control the robot, with good hand-eye coordination and manual dexterity.
- Able to stand at a workstation for an 8-hour shift and do repetitive, precision-focused tasks.
- Meticulous attention to detail and comfort meeting set data collection goals for volume and quality during each shift.
- Available at least 5 days a week on one of the posted shifts: evening (4:00 PM to 12:00 AM PT) or overnight (12:00 AM to 8:00 AM PT).
- No college degree or robotics background is required. Hands-on technical, lab or precision work is the ideal background, and robotics, manufacturing, assembly, lab work or gaming with controllers is listed as nice to have.
Step by step
- 1
Confirm you can commute to San Francisco
The Robot Operator role is fully on-site because you physically drive a real robot arm in the lab. There is no home-office option, so rule this in before you apply.
- 2
Apply on the official Ashby job board
Physical Intelligence posts and takes applications for the Robot Operator role on its own Ashby careers board. Apply there directly rather than through a third-party recruiter; copies of the posting on job aggregators send you back to Ashby.
- 3
Choose a shift and state honest availability
The posting offers an evening shift (4:00 PM to 12:00 AM PT) or an overnight shift (12:00 AM to 8:00 AM PT), on a Mon-Fri, Wed-Sun or Sat-Wed pattern, with a 30-minute lunch and two paid breaks per 8-hour shift. Wider availability, including weekends, makes you easier to schedule.
- 4
Recruiter screen
Expect an initial screen that confirms location, shift availability, physical ability to stand for a full shift, and general fit. Physical Intelligence does not publish its screening script, so treat the exact rounds as undisclosed and prepare to speak plainly about dexterity, reliability, and why the work interests you.
- 5
Interview focused on dexterity, precision, and reliability
Because the job is repeating precise manipulations to a quality bar, the interview is really testing hand-eye coordination, attention to detail, comfort with repetitive metrics-driven work, and whether you will show up for every shift. Bring concrete examples of precise hands-on work (lab, assembly, crafts, gaming). Whether there is a hands-on dexterity trial in the room is not publicly documented.
- 6
Offer and on-site onboarding
If hired, you join as a full-time hourly employee working at least 5 days a week on your chosen shift, with benefits the posting lists as a 401(k), health, dental and vision insurance, paid time off, paid sick time and commuter help. You learn the control system (leading the arm so the robot mirrors you, or guiding it with a controller) and the data volume and quality goals you will be measured against.
Physical Intelligence does not publish a time-to-hire for the Robot Operator role. The posting has been open on its board since August 2026, so applications are taken continuously rather than in rounds.
Robot Operators are full-time hourly employees with benefits, and the posting states one hourly rate for all shifts. The rate in Physical Intelligence's open posting today is on this page, read from its board each day. Confirm the rate and the benefits on the offer before accepting.
Tips to get accepted
- Lead with dexterity, not credentials. Hands-on technical, lab or precision work is the ideal background, gaming with controllers is a listed nice-to-have, and no degree is required.
- Apply directly on the Ashby board rather than through a recruiter.
- Signal reliability above everything. Data quality depends on operators who show up every shift and follow the protocol exactly, and that is the trait every operator posting is really testing for.
- Offer the widest availability you honestly can across the evening and overnight shifts and the Mon-Fri, Wed-Sun and Sat-Wed patterns. Flexibility makes you easier to schedule and to hire.
- Be honest with yourself about standing for a full 8-hour shift and repeating precise motions. It is closer to skilled studio or lab work than a desk job, so do not oversell stamina you do not have.
What isn't public
- Physical Intelligence does not publish its interview steps, the number of rounds, or whether there is an in-person dexterity trial. Those steps here are reasoned from the role, not confirmed by the company.
- No time-to-hire is published.
- The posting lists the kinds of benefits but not their details (plan costs, amount of paid time off, any equity).
- The posting states one rate for every shift; it does not mention a night differential for the overnight shift.
- Shift options can change between versions of the posting (the version read on 28 Oct 2026 offers evening and overnight shifts at 5 days a week), so confirm the schedule with the recruiter.
From the forum
Ask the people doing Physical Intelligence work what only they can tell you.
Frequently asked questions
Is Physical Intelligence a good place to work?
Mostly. Physical Intelligence is a richly funded robot AI lab that hires operators as full-time employees with benefits, but it pays $25 an hour for quota-driven evening or overnight shifts, and no employee rating could be read. The pay is low for San Francisco: a flat $25 an hour, against MIT's $32.44 living wage for a single adult there, for 8-hour standing shifts from 4 PM to midnight or midnight to 8 AM, with data collection targets every shift.
Is Physical Intelligence legit?
Yes. Physical Intelligence (π) is a San Francisco robot AI lab founded in 2024 by former Google DeepMind researchers and Stanford and UC Berkeley academics, including CEO Karol Hausman, Sergey Levine, Chelsea Finn and Lachy Groom. Reports in November 2025 said it raised $600 million at a $5.6 billion value, led by Alphabet's CapitalG. TechCrunch counted about 80 staff in January 2026. In October 2026 it opened its models to other robot makers in a Pi API beta.
Does Physical Intelligence pay?
Yes. Robot foundation-model lab founded by AI academics and former Google DeepMind researchers; its last confirmed round valued it at $5.6 billion. Payouts: Hourly, full-time (at least five days a week), with 401(k), health insurance and paid time off. Robot Operators are full-time hourly employees with benefits, and the posting states one hourly rate for all shifts. The rate in Physical Intelligence's open posting today is on this page, read from its board each day. Confirm the rate and the benefits on the offer before accepting.
How much does Physical Intelligence pay?
$25/hr in checked postings. Source examples collected October 28, 2026. This envelope spans individual advertised offers with different duties, locations or shifts. It is not a company-wide typical wage. See current openings for updated terms.
Is Physical Intelligence worth it?
Physical Intelligence pays $25/hr in checked postings (stated by the employer). It suits people who want paid on-site shift work at a robot lab rather than flexible remote gigs. Worth knowing: Metrics-based role with data-collection goals per shift; 8-hour standing shifts (stated in the posting, not a worker rating).
How does Physical Intelligence pay out?
Hourly, full-time (at least five days a week), with 401(k), health insurance and paid time off.
Where is Physical Intelligence open to workers?
On-site in San Francisco, California; it has also posted technician roles in Fremont.
Do I need a degree or robotics experience to get hired?
No. The role does not require a college degree or a robotics background. It screens for manual dexterity, hand-eye coordination, attention to detail and reliability. Hands-on technical, lab or precision work is the ideal background, and robotics, manufacturing, assembly, lab or controller-gaming experience is a nice-to-have.
What will I actually do on the job?
You teleoperate robotic arms through everyday and assembly tasks: either you lead the movement with your arms and the robot mirrors you, or you guide it with a controller. Example tasks include bagging groceries, sorting cups and utensils, opening jars, folding clothes, installing a light bulb and electronics assembly. You also review and annotate videos of the robot's performance and must meet data volume and quality goals during your shift.
How much does it pay?
It is a full-time hourly job with benefits, and the posting states one rate for every shift. This page shows the rate in Physical Intelligence's open posting today, read from its board each day. Confirm it on the offer before accepting.
Can I do this remotely or from home?
No. It is a fully on-site role in San Francisco, because you operate a physical robot in the lab. If you want remote robot-training work, other employers hire for it. See the teleoperation guide for those paths.
What shifts are available?
Two in the current posting: evening (4:00 PM to 12:00 AM PT) and overnight (12:00 AM to 8:00 AM PT), worked on a Mon-Fri, Wed-Sun or Sat-Wed pattern, at least 5 days a week. Each shift includes a 30-minute lunch and two paid breaks.
Is there a test or trial I need to pass?
Physical Intelligence does not publicly document a formal test for this role (no coding challenge or written exam). The process screens for dexterity, precision, and reliability, most likely in the interview. Whether there is a hands-on dexterity trial is not disclosed, so prepare to demonstrate steady hands and consistency however they ask.
Physical Intelligence versus other employers
Torn between two? Each matchup lines up pay, how hard it is to get in and current openings. Sorted by what an application to either employer is likely worth to you, with a referral lift for an employer that pays us a referral fee; your pay is never affected.
How Physical Intelligence stacks up against similar employers
Data-collection operators who run 1X's humanoid robots in VR to collect training data.
$25 to $36/hr in checked postingsStated by the employerSwing-shift data-collection and model-evaluation operators for AI robotics.
$30 to $35/hr in checked postingsStated by the employerRemote Memory Developers record household-task demonstrations, paid per approved data hour; an in-house team works shifts in Redwood City.
$35/hr in-house; remote $30 to $60 per approved data hourStated by the employerHumanoid robot operators and Helix data creators collecting the demonstrations behind Figure's Helix model.
$30–$65/hrTypical pay, liveSorted by what an application there is likely worth to you, with a referral lift for employers that pay us a referral fee; your pay is never affected.
Hiring for Physical Intelligence? We find the people (you contract them yourself) or we run the project (we deliver the robot data).
Or feature Physical Intelligence's roles for 30 days ($99 for one role, $299 for every open role): a labelled Featured block above the lists and the first roles in the weekly job email.