Human Archive is a research lab backed by Y Combinator focused on modeling human embodied intelligence.
Humans are the most sophisticated biological systems we have ever observed, yet we still do not fully understand ourselves. Research into human physical intelligence — including the human hand, proprioception, and vision — remains largely unsolved. Our mission is to recover human embodied intelligence as a learned model. To achieve this, we build custom hardware products, deploy them globally at scale, and publish research. Today, our data is used for robotics and world modeling, but the broader opportunity is advancing scientific research into intelligence itself.
Founded by Stanford and UC Berkeley researchers, we are lean, deeply technical, and operate at extreme speed, taking on unglamorous and conventionally impossible problems that directly unlock step-function gains in model capability.
The deployment of capable humanoids at scale will permanently redefine human labor. Undesirable physical work will disappear, and human effort will shift toward a new era of abundant creativity.
We are building the infrastructure to accelerate that transition by assembling the Human Archive mafia. You will own meaningful systems from day one and see your work directly impact model capabilities. This is a once-in-a-generation inflection point. If you want to help reshape physical labor and work on problems that matter at civilizational scale, join us.
About the role
We operate a fleet of camera and sensor devices in the field at scale. They generate large volumes of recordings on removable media, and every recording has to reach cloud storage intact, correctly identified, and traceable back to the exact device, card and operator that produced it.
This role owns the technical layer underneath that pipeline. Day-to-day offload is run by an operations lead and their team; you are the person who makes their work possible and correct. You own the ingest stations, the tooling that runs on them, the hardware fleet, quality control, and the standards everyone follows. When something fails in a way the operations team cannot resolve, it comes to you.
It is hands-on. You will spend time at ingest stations, in the quality control area, and at deployment sites, and you will write scripts to make all of it less manual.
It suits someone comfortable with a Linux terminal and a screwdriver in the same afternoon, who finds an unaccounted-for storage card genuinely intolerable.
What you’ll do
Own the ingest pipeline (the operations team runs it)
Build, configure and maintain ingest stations — hardware, operating system, tooling — and replicate that build reliably across locations
Own the ingest tooling in the field: deploy updates, fix what breaks, and carry requirements back to the software team
Define the verification and media lifecycle rules the operations team follows — what must pass before a card is released for reuse, and what happens when it doesn’t
Act as escalation for failed and partial ingests: determine whether the cause is media, device, station, tooling or procedure, and fix the underlying issue rather than the instance
Audit that the process is actually being followed, and close the gaps you find
Own the hardware fleet
Maintain inventory of capture devices, memory cards and drives across locations, with full chain of custody
Run reconciliation and investigate discrepancies to root cause rather than writing them off
Specify and evaluate hardware — cards, readers, hubs, enclosures, drives — and manage returns and warranty claims with suppliers
Plan capacity: how many cards, how many drives, what rotation, with margin
Quality control
Investigate recording failures with controlled tests rather than guesswork; isolate whether a fault sits in the device, the media, the station or the procedure
Decide what gets escalated, repaired or retired, and document the evidence behind the call
Define capture configuration standards and verify device settings in the field
Support the teams
Be the technical escalation point for the operations lead, the quality control team and field operators
Train operators on tooling and hardware, and write procedures a new joiner can follow unsupervised
Travel to deployment sites as needed
Report
Produce recording yield, data loss and operator performance reporting
Be able to say whether a bad number reflects a real failure or an artefact of how it was measured
Automate
Write Python and shell scripts to remove manual steps from the workflow
Work with the software team on tooling requirements, and file precise bug reports when the software gets in the way
A technical background — a degree in engineering or computer science, or equivalent hands-on experience. This is an engineering role that happens to involve hardware and field work, not a coordination role with a technical veneer
Comfort in Linux: mounting and formatting media, filesystems, permissions, scheduled jobs, reading logs
Working Python — enough to script a batch job, walk a directory tree, or call an API
Fluent use of AI tools — Claude, Claude Code, Cowork or equivalent — as part of how you work day to day. We expect you to use them to write scripts, investigate failures and draft procedures faster than you would alone, and to know where their output needs checking before it reaches a station or an operator
Hardware troubleshooting instinct: forming a hypothesis, testing it, eliminating causes one at a time
Real care about data integrity and chain of custody, and the discipline to follow a verification step even when it is inconvenient
Clear written English — you will write procedures other people depend on
Willingness to work with physical equipment and travel to sites
Comfort in an environment where the process may not exist yet and you are expected to define it
Experience with cameras or sensor hardware
SQL, or experience querying an internal API for reporting
Storage systems — RAID, ZFS, network-attached storage administration
Prior work in data collection, field operations, logistics or laboratory operations
Barcode or asset-tracking systems
