The team map
Apptronik's 23-person engineering team has iterated through nearly 80 custom electric actuator designs to put Apollo, its humanoid robot, onto Mercedes-Benz factory floors and GXO Logistics warehouses in 2025. The company recruits mechanical, electrical, and software engineers who prove they can deliver working hardware under constraints, not just academic credentials, because actuation work and a deepening AI partnership with Google DeepMind around Gemini 2.0 define three core disciplines: mechanical engineers who design for high-cycle durability, electrical engineers who pack compute and sensing into tight electromechanical packages, and software engineers who close the loop between perception, planning, and low-level control.
Sebastien Cotton, Vice President of Hardware Engineering and Chief Engineer, oversees mechanical and electrical. Chirag S. leads software. Amit Singla runs Applied AI. Lomesh Agarwal directs Perception, Navigation and Autonomy, a senior director role signaling onboard autonomy as a dedicated pillar. Maneesh Jain runs Technical Program Management, coordinating hardware, software, and the supply chain work Cardenas calls "vertical integration we'll earn over time."
Daniel Chu, Chief Product Officer, scaled teams at Waymo, 23andMe, and Google. Kathryn Marinaro leads User Experience; Shaun Wellens heads Industrial Design. The careers page and Zero G Talent postings show the mix: Principal Electrical Engineers for compute and sensing, Staff Electrical Engineers for compute subsystems, Senior Perception Learning Engineers for SLAM, Staff Product Managers for Applied AI, and a Director of Product Management for Software in Mountain View near the DeepMind partnership.
Cardenas says Apptronik has "one of the best actuation teams in the world right here in Austin." The team sits inside Cotton's hardware org but runs with enough autonomy to iterate through dozens of variants.
Software hiring splits between the autonomy stack (perception, navigation, SLAM under Agarwal) and the robot-level control stack (real-time controllers, safety, simulation under Chirag S.). The DeepMind collaboration adds engineers who integrate Gemini 2.0 into the robot's reasoning layer, handled by Singla's Applied AI group.
A small customer success team (three people) and a field services function under Kevin Garell, SVP of Global Services & Support, pull from the same engineering pool: technicians and application engineers who deploy, monitor, and maintain Apollo at pilot sites. Steve O'Dea, leading Engineering, Manufacturing, Field Service, and IT, calls his mandate fusing "world-class engineering with disciplined operational execution so Apollo delivers immediate customer value on factory floors, in warehouses, and beyond."
Pay runs above the robotics median
Apptronik pays above the robotics-startup median, driven by the capital intensity of humanoid development and a $935M Series A closed in February 2026. The cleanest anchor is Zero G Talent's live feed: Zero G Talent's data shows a band of $142k–$321k, median $235k across 13 roles posted since late 2025, drawn from open requisitions.
Third-party sources report lower medians ($155k to $218k) because public postings skew toward individual contributors, while the board captures staff, principal, and director requisitions that rarely hit generic job sites.
| Role (as posted on Zero G Talent) | Location | Salary Range (USD/year) |
|---|---|---|
| Director of Product Management, Software | Mountain View, CA | Zero G Talent found $315,000 – $350,000 |
| Principal Electrical Engineer – Compute and Sensing | Sunnyvale, CA | Zero G Talent's figures put $295,000 – $330,000 |
| Staff Technical Program Manager | Sunnyvale, CA | $250,000 – $285,000 |
| Staff Product Manager, Applied AI | Mountain View, CA | $215,000 – $245,000 |
| Staff Electrical Engineer – Compute | Sunnyvale, CA | $210,000 – $235,000 |
| Senior Perception Learning Engineer – SLAM | Sunnyvale, CA | $190,000 – $235,000 |
All six roles are Bay Area–based, inflating numbers relative to Austin. No formal geographic differential is published. Austin candidates should expect a downward shift at each level, based on comparable titles at other robotics firms.
Equity matters: the raise implies a post-money valuation north of $5 billion, so even modest grants carry paper upside. The company doesn't publish strike prices or refresh schedules; ask for the current 409A and typical grant size at offer. Benefits follow the startup template: medical, dental, vision, 401(k) match, flexible PTO, hardware stipend. The differentiator is access — robot time, machine-shop priority, a test cell running 24/7. That access is what the top of the band buys. The pay reflects the constraint: working hardware, shipped on schedule, in an environment that doesn't forgive simulation-only validation.
If you're negotiating, lead with the board band — it's the only dataset the recruiter can't dismiss as stale or aggregated. Ask where the role sits relative to the $142k–$321k span, then press for the equity grant size and the 409A date. The numbers are public; the conversation is not.
Inside the interview loop
The interview pipeline runs five to six rounds over three to five weeks, though deep robotics backgrounds and strong referrals can compress it to two or three. The process filters for engineers who operate across the full stack: perception, planning, controls, and the hardware interfaces that bind them — not specialists in one layer.
The first gate: a 30- to 45-minute recruiter screen covering background, motivation, and alignment with five values: Curiosity, Humility, Integrity, Passion, Creativity. Candidates who can't articulate why they want to build general-purpose humanoids, not industrial arms or simulation-only research, stall here. The recruiter also verifies minimums: professional C++ and Python, hands-on ROS 1/2, a track record shipping software on physical robots.
A technical assessment follows, often a take-home: sensor fusion, trajectory optimization, or a controls simulation in a few days. Evaluators weigh code clarity, architectural decisions, and handling of real-world constraints: latency, sensor noise, compute budgets. Research-grade solutions that ignore deployment realities signal a candidate who hasn't operated in product.
The onsite loop, in Austin or Sunnyvale, runs three to four back-to-back sessions. One covers robotics fundamentals: kinematics, dynamics, state estimation, control theory on legged platforms. A second probes ROS architecture: node design, real-time partitioning, debugging across compute boards. A third tests C++ fluency where it matters: template metaprogramming for zero-cost abstractions, lock-free structures for inter-thread communication, memory management in long-running processes. Python gets a separate pass for tooling, data pipelines, and rapid prototyping.
A system-design interview closes the technical evaluation. Judges look for problem decomposition, module interfaces, compute allocation, and hardware verification plans. Strongest candidates cite specific failures debugged on real robots.
The behavioral round evaluates cross-functional collaboration. Apptronik's teams blend mechanical, electrical, and software engineers daily; candidates who describe hardware colleagues as "blockers" or "ticket queues" rarely advance. Interviewers look for shared ownership — instances where the candidate instrumented a board bring-up, co-designed a test fixture, or rewrote a driver to expose telemetry the controls team needed.
Screening follows a two-step filter: minimum qualifications first. If no, stop. If yes, preferred qualifications: prior humanoid or legged-robot experience, open-source robotics contributions, demonstrated ability to lead a subsystem from concept to fielded hardware. The acceptance rate for qualified applicants is 3–6%, reflecting the narrow intersection of deep robotics software skill and hardware-integration pragmatism.
Feedback comes through recruiting. Hybrid schedules now dominate software roles, with onsite days tied to hardware milestones. What separates hires from near-misses is not algorithmic cleverness on a whiteboard. It is the ability to articulate, with specificity, how a software decision propagates through actuators, sensors, and mechanical structure, and how to verify that propagation before the robot steps onto the lab floor. That same verification mindset shapes the factory floor where those robots are built.
The factory floor
Apptronik's footprint traces a research spinout turning production-scale. The anchor is Austin, where the company began in the Human Centered Robotics Lab at UT before spinning out in 2016. As of June 2026, headquarters occupies a nearly 90,000-square-foot facility in North Austin called Robot Park, where fleets of Apollo 2 robots, bipedal and wheeled, run logistics, manufacturing, and retail tasks to generate the embodied AI training data the company treats as its second product line. "We have a factory that produces robots, we also have a factory that produces data," CEO Jeff Cardenas said in June 2026, calling Robot Park the engine for building production-ready models through continuous real-world operation.
In early 2026 Apptronik leased an additional 59,000-square-foot warehouse across the street, pushing the Austin campus toward 150,000 square feet for roughly 300 employees. The expansion adds assembly cells, hardware-in-the-loop validation, and repeated manipulation cycles that turn lab-grade actuators into field-reliable subsystems. Engineers share floor space with the robots they're iterating on, shortening the loop from machine-shop design change to test-cell validation.
California is smaller but deliberate. Roles in Sunnyvale and Mountain View (Senior Perception Learning Engineer (SLAM), Staff Electrical Engineer (Compute), Principal Electrical Engineer (Compute and Sensing), Director of Product Management) map to perception, compute, and real-time controls that complement Austin's actuation focus. These aren't sales outposts; job descriptions demand hands-on bring-up of custom compute and sensor suites, requiring bench space, oscilloscopes, motion-capture volume. Sunnyvale functions as a deep-tech satellite where the software-hardware boundary gets stress-tested daily.
The Robot Park concept ties the sites together. Austin anchors a growing global network at customer and partner sites. Each park extends data collection: robots in Mercedes-Benz production lines, Google DeepMind research environments, and other deployments feed telemetry back to core models. For engineers, the "lab" is a distributed fleet. The toolchain spans simulation pipelines, hardware-in-the-loop rigs, and Robot Park's physical test courses that expose edge cases no simulator catches.
The facilities reflect the company's bet: reliability emerges from volume of real-world operation, not cleaner demos. A machine shop turning a revised actuator housing in hours, a test cell running endurance cycles, a warehouse bay where robots share a workspace — these are the instruments. Engineers write code that ships to metal the same week, then watch it fail or succeed on a robot nearby. That proximity between design authority and physical consequence defines the work environment and the profile of engineers who last in it.
The profile that lasts
Apptronik's communications and employee accounts agree: the profile is engineers who treat robotics as systems integration, not isolated disciplines. The company builds humanoids that perceive, plan, and act in unstructured environments: warehouses, factory lines, eventually hospitals and homes. That scope filters for people who've debugged a perception stack while the actuator team waits on a torque curve, or rewritten a motion planner because the mechanical design changed overnight.
Mission alignment screens first. Company materials frame the work around "robots for humans" and "Man + Machine", taking on dull, dirty, dangerous tasks so people focus on higher-value work. Cardenas calls the humanoid race "the space race of our time." Candidates who treat that as marketing leave. Those who stay cite clarity of purpose: Apollo reduces physical strain, improves safety, addresses labor shortages in manufacturing, logistics, warehousing. Mercedes-Benz and NASA partnerships mean robots ship to real workflows, not demos. Engineers who need that validation (seeing their code lift a tote on a Mercedes line) thrive. Research purists don't.
Technical profile outweighs pedigree. The careers page and engineering blog emphasize end-to-end exposure across controls, perception, manipulation, test/validation. A 355-person team with a Robot Park in Austin means mechanical, electrical, and software engineers share lab space daily. Cross-disciplinary debugging is the default. Employees describe high learning velocity through hands-on lab work and visible partner projects. Success traits: comfort reading schematics while writing ROS 2 nodes, willingness to instrument a test cell before a pilot shipment, ability to explain a control-loop instability to a mechanical designer who doesn't write code.
Cultural signals reinforce the bar. The company promotes from within, runs quarterly engagement surveys, uses OKRs. Open office, in-person all-hands, team-based planning signal communication isn't optional. Open-door policy and hybrid schedule (remote Mondays or Fridays at manager discretion) give autonomy with accountability. BuiltIn's values read generic until practice: humility when a perception engineer admits the sensor suite has limits; integrity when a test lead flags a safety interlock failure before a customer visit.
Pace is the final filter. The "space race" framing isn't rhetorical — major Apollo upgrades shipped in 2025, and Disruptor 50 recognition reflects pressure to iterate faster than U.S. and Chinese competitors. Glassdoor shows 61% would recommend the company, with work-life balance at 3.5/5 and culture at 3.7/5 — the mission carries people through crunch, but the balance score warns the pace isn't sustainable for everyone. Engineers who last treat a shifting spec as a puzzle, not a grievance. The robot doesn't care about your resume — only whether the hardware holds.
Working in frontier tech? Zero G Talent tracks the openings: see every open Apptronik role, browse frontier tech jobs, the companies hiring, and the people building the field.