B
Principal Robotics Engineer
Accepting applicationsBoolSi · San Francisco Bay Area
Full-Time Senior ARMC++FPGA
Posted
3d ago
Category
Design
Experience
Senior
Country
United States
00About
We know compilers. We're hiring the person who knows robotics.
Robots running on embedded ARM cores are often compute-bound: the CPU budget caps what they can perceive and how fast they can react. The usual fix is a GPU. Many robots can't take one: a few-watt power budget, hard latency bounds, a fanless enclosure, or a certification path.
An FPGA fits those constraints, and MPSoCs like Zynq and Kria put one on the same chip as the ARM cores. Few robotics teams use it, because programming it has meant hiring chip designers. BoolSi removes that requirement: our compiler turns C, C++, and Rust functions into FPGA accelerators that behave identically to the original code. We've raised $7.6M and are a team of 12 engineers and scientists.
Boston, MA, on-site. The robots are in the lab.
Reports to the CEO.
$180K–$250K plus meaningful equity.
Travel around 10% for the first six months, 20–30% after.
01The role
Three overlapping phases.
The core of the job is telling us which workloads to go after: where robotics systems are actually compute-bound, and which of those bottlenecks the market will pay to remove. Your judgment drives what we dogfood, what we demo, and what we pitch.
Months 0–3: User zero. Run realistic robotics workloads through the platform before any customer does. Candidates so far: LiDAR processing, visual SLAM front-ends, MPC and other high-rate control loops, large-state sensor fusion such as EKFs. Some will turn out bandwidth-bound rather than compute-bound; finding out fast is the point. Define how we benchmark: latency, jitter, power, and control rate against the ARM baseline and, where a customer would consider one, a Jetson-class GPU. You will get blocked. A compiler engineer is paired with you to unblock you, and every block becomes a precise bug report or feature request. By month 3 the platform is measurably better because of you.
Months 2–6: Physical demo. Build a reference demo on an off-the-shelf ROS 2 robot with a Kria board: sensor and actuator wiring, system integration, the accelerated workload in the loop. It runs the same workload with and without BoolSi acceleration, and the accelerated robot does something the ARM cores alone cannot. It ships within 6 months, runs reliably at a trade show, and makes the point in seconds. One demo, done well.
Month 6+: Field work. Lead customer proofs of concept: find the hot function, port it, benchmark it, present the results. Run the demo at trade shows and conferences, and write and speak publicly about the work. Feed what you learn into the roadmap. A second robotics engineer is budgeted at month 6, and a field application engineering team follows the first paid POC; you hire both. By month 12: two POCs delivered with benchmark reports customers cite internally, and a documented field playbook the team runs.
02Trajectory
Where this goes.
Two good outcomes at month 12: you lead the field team you started hiring, or you stay principal robotics engineer and we hire someone to run it beside you. We decide together around month 9. Either way you own which workloads we build for.
03Fit
What we're looking for.
Staff-level experience (roughly 8+ years) in robotics or embedded software, with software shipped on physical robots, whether production systems or serious research platforms.
Strong C/C++ on embedded Linux, with experience profiling and reasoning about real-time constraints, latency, and jitter.
Informed opinions about where robotics compute budgets hurt today, grounded in systems you've worked on, and the judgment to tell technically interesting problems apart from commercially important ones.
Hands-on bench competence: wiring, soldering, sensor integration, keeping a physical platform running.
Demonstrated communication ability: conference talks, teaching, technical writing, or prior customer-facing engineering.
Working ROS 2 fluency for the demos. Many customers won't use it, so production experience on non-ROS stacks counts as much.
04Not required
Explicitly not required.
FPGA or HDL experience. Our product automates the hardware design, and your first few months double as a test of that claim. Where it fails, that is a bug report.
Mechanical design. Demos are built on off-the-shelf platforms that you integrate and program.
Kernel or BSP work. Our platform team owns the Kria images, device trees, and drivers. You get a board that boots; you make it a robot.
Management experience. If you take the team-lead path, you build your first team with the CEO's help.
Show more Show less
We know compilers. We're hiring the person who knows robotics.
Robots running on embedded ARM cores are often compute-bound: the CPU budget caps what they can perceive and how fast they can react. The usual fix is a GPU. Many robots can't take one: a few-watt power budget, hard latency bounds, a fanless enclosure, or a certification path.
An FPGA fits those constraints, and MPSoCs like Zynq and Kria put one on the same chip as the ARM cores. Few robotics teams use it, because programming it has meant hiring chip designers. BoolSi removes that requirement: our compiler turns C, C++, and Rust functions into FPGA accelerators that behave identically to the original code. We've raised $7.6M and are a team of 12 engineers and scientists.
Boston, MA, on-site. The robots are in the lab.
Reports to the CEO.
$180K–$250K plus meaningful equity.
Travel around 10% for the first six months, 20–30% after.
01The role
Three overlapping phases.
The core of the job is telling us which workloads to go after: where robotics systems are actually compute-bound, and which of those bottlenecks the market will pay to remove. Your judgment drives what we dogfood, what we demo, and what we pitch.
Months 0–3: User zero. Run realistic robotics workloads through the platform before any customer does. Candidates so far: LiDAR processing, visual SLAM front-ends, MPC and other high-rate control loops, large-state sensor fusion such as EKFs. Some will turn out bandwidth-bound rather than compute-bound; finding out fast is the point. Define how we benchmark: latency, jitter, power, and control rate against the ARM baseline and, where a customer would consider one, a Jetson-class GPU. You will get blocked. A compiler engineer is paired with you to unblock you, and every block becomes a precise bug report or feature request. By month 3 the platform is measurably better because of you.
Months 2–6: Physical demo. Build a reference demo on an off-the-shelf ROS 2 robot with a Kria board: sensor and actuator wiring, system integration, the accelerated workload in the loop. It runs the same workload with and without BoolSi acceleration, and the accelerated robot does something the ARM cores alone cannot. It ships within 6 months, runs reliably at a trade show, and makes the point in seconds. One demo, done well.
Month 6+: Field work. Lead customer proofs of concept: find the hot function, port it, benchmark it, present the results. Run the demo at trade shows and conferences, and write and speak publicly about the work. Feed what you learn into the roadmap. A second robotics engineer is budgeted at month 6, and a field application engineering team follows the first paid POC; you hire both. By month 12: two POCs delivered with benchmark reports customers cite internally, and a documented field playbook the team runs.
02Trajectory
Where this goes.
Two good outcomes at month 12: you lead the field team you started hiring, or you stay principal robotics engineer and we hire someone to run it beside you. We decide together around month 9. Either way you own which workloads we build for.
03Fit
What we're looking for.
Staff-level experience (roughly 8+ years) in robotics or embedded software, with software shipped on physical robots, whether production systems or serious research platforms.
Strong C/C++ on embedded Linux, with experience profiling and reasoning about real-time constraints, latency, and jitter.
Informed opinions about where robotics compute budgets hurt today, grounded in systems you've worked on, and the judgment to tell technically interesting problems apart from commercially important ones.
Hands-on bench competence: wiring, soldering, sensor integration, keeping a physical platform running.
Demonstrated communication ability: conference talks, teaching, technical writing, or prior customer-facing engineering.
Working ROS 2 fluency for the demos. Many customers won't use it, so production experience on non-ROS stacks counts as much.
04Not required
Explicitly not required.
FPGA or HDL experience. Our product automates the hardware design, and your first few months double as a test of that claim. Where it fails, that is a bug report.
Mechanical design. Demos are built on off-the-shelf platforms that you integrate and program.
Kernel or BSP work. Our platform team owns the Kria images, device trees, and drivers. You get a board that boots; you make it a robot.
Management experience. If you take the team-lead path, you build your first team with the CEO's help.
Show more Show less