The hiring footprint
Agility Robotics has deployed Digit at GXO Logistics and Schaeffler sites with a roughly 36-hour integration window — a claim that only holds because the company hires across robotics, hardware, software, and operations as a single deployment operation. Founded in 2015 before the transformer architecture existed, Agility splits across three primary centers: Salem, Oregon (headquarters and primary manufacturing), Fremont, California (engineering and AI), and Pittsburgh, Pennsylvania (advanced research), with a growing remote tier for senior AI and software roles.
First-party board data (Zero G Talent's board data) shows 59 salaried postings across a $98k–$337k band (median $226k). The distribution tells the real story.
| Role | Location type | Salary band (USD/year) |
|---|---|---|
| Senior Director, AI | Remote | 257,000 – 402,000 |
| Vice President, Controller | Hybrid (Fremont, CA) | 285,000 – 370,000 |
| Senior Manager, AI Innovation | Remote | 268,000 – 364,000 |
| Lead Data Scientist, Robotics | Remote | 218,000 – 340,000 |
| Senior Staff Software Engineer, Calibration | Remote | 218,000 – 340,000 |
| Staff AI Research Engineer | Hybrid (any office) | 216,000 – 338,000 |
Three patterns emerge. AI and data-science leadership commands the widest bands — the Senior Director, AI role spans $145,000, signaling that outcomes vary sharply with impact. Remote eligibility is standard for senior software and research positions; only the finance leadership role and the hybrid research engineer require office presence. Calibration and controls engineering (core to making Digit reliable in customer facilities) sits at the $218,000–$340,000 tier, on par with AI research staff. That parity tells you where the technical risk lives: getting a biped to repeatably unload Spanx boxes onto a conveyor at GXO is worth as much as advancing the foundation model.
The AI team owns the "digital semantic intelligence" layer, using large language models to author workflows in the field, while maintaining an architecture that can swap foundation models underneath. The company describes itself as AI-agnostic; the hiring reflects that pragmatism. Below the AI layer, the software stack splits into calibration, simulation, and fleet autonomy. Calibration is the hidden lever: every Digit that walks into a new facility must self-calibrate against that building's geometry, lighting, and conveyor heights within the same day-and-a-half deployment window. The data scientist role feeds the same loop, turning field telemetry into model updates. These are not backend roles; a regression means a robot stops at a customer site.
Hardware and manufacturing sit in Salem. The VP Controller role at $285k–$370k (hybrid Fremont) signals a finance function that understands hardware unit economics. Deployment operations, the team that actually installs robots at GXO and Schaeffler sites, recruits from field service and systems integration backgrounds. They need engineers who can read a warehouse safety plan, negotiate with a facilities manager, and debug a perception stack before the shift starts.
Safety engineering cuts across all of it. The company works with the U.S. government on humanoid regulatory requirements and cites Amazon's safety bar as the highest they've encountered. That constraint shapes hiring: every team (AI, software, hardware, deployment) screens for systems integration ability and execution rigor. The org chart is flat by design; the "who gets hired" question resolves to people who can own a subsystem from lab to loading dock without handoffs.
Equity and variable components are not broken out in the board data. The board data captures 59 salaried postings; hourly and contract roles for technicians, operators, and field deployment staff are not included in the median.
Inside the hiring filter
Agility Robotics does not publish a detailed interview playbook. What exists instead is a pattern visible in the roles the company posts, the technical demands of its product, and the few public statements from leadership about what it takes to ship a humanoid into a warehouse.
The board data shows the company recruiting for roles that span AI research, calibration software, data science, and senior engineering leadership, all remote or hybrid across its three sites. That distribution alone signals a process that must evaluate cross-disciplinary fluency.
The only public window into the company's hiring philosophy comes from a Bloomberg Live interview (yLKl0Stuy54) with leadership. In that conversation, the speaker emphasizes two non-negotiables that map directly to hiring filters: safety and systems integration. "The biggest bar is safety. You have to operate these safely. If you don't, you won't be allowed in the customer's facilities," said in the Bloomberg Live interview. That constraint means any engineering hire — whether in perception, planning, or low-level control — must demonstrate an ability to reason about failure modes in unstructured environments, not just benchmark performance on clean datasets. The same interview notes the company is "working with the U.S. government to help write the humanoid regulatory requirements."
The labor-gap framing — "about a million unfilled jobs in the US alone… no one wants those jobs" — also shapes the operator and field-automation hiring bar. Agility deploys Digit at GXO Logistics sites under commercial contracts. Digit has been "working in a commercial environment for a year" at GXO, per the same interview. Anyone hired into field roles inherits that context.
What can be inferred from the role slate and the product's maturity is a hiring process that screens for three traits:
Hardware-software closure. The board lists calibration, AI research, and data science as distinct senior tracks. The calibration role explicitly demands this; the AI roles implicitly require it.
Safety-as-a-system mindset. The statement that safety is "the biggest bar" and the company's involvement in writing regulatory standards mean interviews probe for experience with functional safety, risk assessment, and the documentation trail that lets a customer's EHS team sign off.
Deployment durability. The 10-to-1 work-to-charge ratio target mentioned in the interview ("charge for 10 minutes, operate for 100 minutes") is a systems target that spans battery management, motion planning efficiency, and thermal design.
The research contains no leaked interview packet, Glassdoor composite, or careers-page flowchart. What it does contain is a company that has moved from lab prototype to paid deployment at a major 3PL, is engaging regulators, and is hiring senior individual contributors across three geographies to scale that deployment. The roles on the board, and the product in the warehouse, are the clearest spec the company has published.
Three work environments
Agility Robotics operates across a distributed footprint that reflects its dual identity as a robotics developer and a deployment partner. Engineering centers cluster in three locations (Salem, Oregon; Fremont, California; and Pittsburgh, Pennsylvania), each appearing in first-party job postings as hybrid-office anchors. Salem carries the historical weight: the company spun out of Oregon State University in 2015. Fremont situates the team inside the Bay Area. Pittsburgh adds Carnegie Mellon's robotics ecosystem.
Development proceeds inside work cells — waist-high enclosures that keep robots and humans physically separated during testing. As the company's leadership has stated, "All of the robots have to be within work cells, so they're about waist high. Humans aren't in there, robots aren't outside." The cells serve as the primary integration spaces where mechanical, electrical, and software stacks converge on a single Digit unit. Calibration, gait validation, and payload testing all occur inside these bounded volumes.
A transition is underway. The company has announced plans to demonstrate cooperatively safe robots — Digit units permitted to leave the cell and operate in close proximity to people. That shift redefines the lab environment: perception stacks, emergency-stop architectures, and human-robot interaction protocols move from simulation into the same physical space where engineers work. The safety bar is set by the most demanding customer. Amazon's requirements are described as "probably the highest," and Agility is simultaneously engaged with the U.S. government on that framework.
Deployment sites constitute a second class of work environment. At GXO, a third-party logistics provider, Digits have been moving Spanx inventory in a live fulfillment center for over a year. The integration timeline is compressed: "When we come to a new facility, we can be up and running about a day, a day and a half." The warehouse itself becomes an extension of the test cell: narrow aisles, overhead shelving, and the same mixed-automation "islands" that characterize the broader logistics landscape.
The third environment is remote. Senior AI, software, and data roles are listed as fully remote, reflecting a split between the hardware-adjacent work that demands physical presence and the model-training, simulation, and workflow-authoring work that does not. The company uses large language models (Gemini, as of mid-2025) for semantic intelligence and workflow authoring, and it has run reinforcement-learning pipelines for roughly three years. That compute-heavy stack can be developed from anywhere; the hardware loop cannot.
For a candidate, the practical takeaway is geographic intentionality. Roles tied to mechanical integration, electrical bring-up, safety certification, and field deployment will require regular time in Salem, Fremont, or Pittsburgh, and periodic travel to customer warehouses. Roles centered on foundation-model integration, simulation infrastructure, or data pipelines can be executed remotely but still demand fluency with the constraints of the physical robot.
The profile that lasts
The people who stay and advance at Agility Robotics share a recognizable profile: they treat the robot as a system, not a collection of subsystems, and they measure progress by whether Digit can show up at a customer site and do useful work for a full shift.
Safety obsession is the baseline. Deployment speed is the next filter. The claim of a day-and-a-half integration only holds if the team arriving on site includes engineers who can integrate perception, planning, and manipulation stacks against a warehouse management system they've never seen before.
Pragmatic design judgment separates the hires who last from the ones who churn. The leadership team described stripping a five-finger hand down to a simpler end-effector because "you don't use five fingers that often... You need dexterity and fine usability, but not so much five fingers." They chose a "very calm color" for Digit: "not the robot of our nightmares or, you know, sci fi movies." Engineers who argue for elegance over deployability tend to self-select out; the ones who stay treat every added degree of freedom as a liability until proven otherwise.
AI fluency is now table stakes, but the flavor matters. The company has run reinforcement learning on Digit for roughly three years and currently uses Gemini as a foundation model while remaining "AI agnostic... we built the robot to, you know, we can swap out AI models underneath." They use LLMs for "digital semantic intelligence" for that purpose.
Customer-facing grit is non-negotiable. The reference deployments (GXO unloading Spanx totes onto conveyors, Schaeffler loading greasy machine parts into washers) are dirty, repetitive, and physically constrained. "They have to wash machine parts. They're very greasy and they get thrown into a metal bin and Digit takes them and loads those into the washer."
Finally, the compensation data reinforces the profile. The board's 59 salaried roles span that band, with AI and calibration specialists at the top. That spread rewards depth in a hard discipline plus the breadth to integrate it. The warehouse floor does not grade on curve; it grades on whether the robot shows up tomorrow.
Working in frontier tech? Zero G Talent tracks the openings: see every open Agility Robotics role, browse frontier tech jobs, the companies hiring, and the people building the field.