Careers at SWORT
SWORT works with engineers and researchers across embedded systems, avionics, controls, sensing, simulation, and aerospace electronics.
What the work is
Firmware that runs deterministically on constrained targets and still makes decisions worth trusting. Register-level bring-up, interrupt and DMA architecture, RTOS scheduling, and the interfaces between hardware and the software that depends on it.
Boards that survive contact with a real vehicle. Mixed-signal layout, power integrity, grounding and decoupling strategy, connector and harness decisions, and the DFM work that separates a schematic from something manufacturable.
Deciding what a system knows when its sensors disagree. Residual analysis, fault isolation, hardware-in-the-loop testing, and building the evidence that a design behaves as specified when inputs are noisy or partly wrong.
How work happens
Work is organised per program, not per department. You will usually be one of a small number of engineers on a given effort, and you will know what the rest of it depends on.
Subsystems are owned end to end, from requirement through bring-up to verification. There is no layer between you and the consequences of a design decision.
Some work is internal research on sensor trust and system integrity. Some is scoped customer engineering with fixed deliverables. Both exist, and they are staffed differently.
Most engagements today are contract, project-based, or research collaborations rather than full-time employment. We say which one applies before any technical conversation starts.
Design reasoning is written down. Firmware and layouts are reviewed, and a rejection comes with a rationale. Deliverables include the schematics, source, and test records, not just the artifact.
Work is remote unless a program needs hands on hardware, in which case bench time is arranged around the DC and Maryland area. Requirements are stated per engagement, not assumed.
// Where bench work happens
Hardware work takes place at instrumented benches with oscilloscopes, logic analysis, and power measurement available. Some verification runs in shared technical facilities rather than premises owned by SWORT, and which applies is stated per engagement.
Opportunities
These are the disciplines we work in and take introductions for. Status is stated honestly for each one.
Confirmed active opening a funded role being actively filled. Project-dependent opens when a specific program is funded. Rolling interest reviewed continuously, with no committed start date.
No category below is a confirmed active opening today. Nothing here is a guaranteed vacancy, and we will tell you which of the three applies before taking your time.
Firmware in C on STM32-class targets: register-level peripheral work, interrupt and DMA architecture, RTOS task design, and board bring-up through sensor validation.
Schematic capture and layout in KiCad or Altium: mixed-signal boards, controlled impedance, power integrity, grounding and decoupling strategy, and DFM-clean output.
Attitude and state estimation, filter design and tuning, and control law implementation that has to run on an embedded target rather than only in a notebook.
Hardware-in-the-loop rigs, fault injection, sensor and disturbance models, and the test plans that decide whether a system has actually been verified or only exercised.
Sensor trust, residual consistency, and fault isolation methods. Collaboration with researchers and students on publishable work, scoped case by case.
Requirements capture, verification planning, documentation structure, and keeping engineering records traceable across a program's revisions and deliverables.
What we look at
We read submissions looking for specific things. These are them.
Something that ran outside a lab bench: what it did, what constraints it operated under, and which parts were yours rather than the team's.
A repository, a schematic set, a report, or a writeup we can actually read. Private work is fine if you can describe it at a level you are permitted to share.
Why you chose an approach, what you rejected, and what the tradeoff cost you. A defended decision tells us more than a finished product.
How you convinced yourself the thing worked. Measurements, captures, test plans, and the failures you found before someone else did.
Knowing what you can and cannot disclose from prior employers or programs, and saying so plainly instead of oversharing or going vague.
Stating what a design does not cover, what was not tested, and where your own confidence ends. This is the single strongest signal on this list.
Submit technical interest
Six fields. Optional detail is there if you want it, and skipping it costs you nothing.
Every submission is read by an engineer. If your background fits current or upcoming work, we will reach out. If we do not, it means there is no honest match right now rather than a judgement on your work.