Embedded Software Engineer, Airships to Cloud

## About HyLight Power lines, gas pipelines and railways are still inspected the way they were 50 years ago: by helicopter. Expensive, noisy, and up to a tonne of CO₂ per flight hour. We built the alternative. The HyLighter is a 12-metre autonomous airship running on hydrogen. It flies for up to 10 hours, almost silently, and spots a 1 mm defect from 25 metres up. But the airship is only half the story. Every flight brings back massive amounts of LiDAR, imagery and methane data, and that's where the value is. Our AI turns it into what operators act on: 50+ types of defects detected automatically, and leaks pinpointed across their network. It's not a prototype. HyLighters already fly missions for France's largest energy and rail operators and have flown over the Palace of Versailles. Backed by Y Combinator, we just raised €13M to go from 4 airships to 15. **Working at HyLight** We design, build and fly every HyLighter, and turn its data into insight, all under one roof on a former airbase. Which means: - **You own it all the way.** At a big aerospace company, you'd own a small piece of one subsystem. Here, you design it, test it, stand on the airfield when it flies, and see the client act on the data it brings back. - **Weeks, not years.** Ideas go from whiteboard to flight test in weeks. - **Responsibility from day one.** The founders' average age is 27. Nobody here waits for permission. We don't care about titles or ticking every box. We care about what you build and the perspective you bring, so if your profile is different, apply anyway. ## Job Summary Every HyLighter flight produces a lot of data: LiDAR, imagery, methane data. That data only has value if it gets out of the airship, complete and on time. The embedded software is where that starts, and you will own two pieces of it: - **Payload.** The software on board the airship that captures the data. - **Ops Upload tool.** The software in our field trucks that offloads the data after each operation and sends it to the cloud. Today both work, and they support real missions for real customers. But we are going from 4 airships to 15, and "works" is not enough. The fleet will fly more, in more places, with operators who did not write the code. As an Embedded Software Engineer, you will make this software scalable, robust, and a pleasure to use for the operator on the field. The goal is simple to state and hard to reach: seamless operations, and data flowing from the whole fleet into our post-processing platform, both during operations and while the airships are parked in their trucks. ## Mission Within the data team, working closely with the Head of Data and hand in hand with operations, you will: **Make Payload bulletproof** - **Reliable capture.** Complete, valid data on every flight, and problems caught before takeoff, not after landing. - **Self-healing.** When something fails, the system recovers on its own, without the operator having to step in. - **Tested like hardware.** Changes are validated on real devices before they fly. - **Fleet-ready.** The same setup and the same updates across 15 airships, with no hand-tuning for each one. **Make the Ops Upload tool foolproof** - **Fast and safe offload.** The operator plugs in, and the data is offloaded and verified without needing to understand how. - **Upload that always finishes.** Even with poor connectivity in the field. - **Operator-friendly.** Clear status, clear errors, as few manual steps as possible. - **Visible.** We know in real time where every file is, during operations and while the airships are parked in their trucks. **Connect it to the platform** - **Own the handover.** Define with the rest of the data team what leaves the airship and the trucks, so every flight is traceable from capture to customer insight. - **Close the loop.** When something breaks in the chain, find the root cause across hardware, network and software, and make sure it doesn't happen twice. ## Personal Note from Louis (Head of Data) "The airship only matters if its data arrives. Right now our embedded software does the job thanks to a small team that knows every corner of it. I'm looking for someone who wants to take it from 'works for the people who wrote it' to 'works for 15 airships and for every operator on the team, every day'. You'll have real autonomy, your code will run on a machine you can watch fly, and you'll see the result in the field the same week." ## Profile Sought This role is embedded first. You may come from firmware, robotics or a systems background, and you want to see your software run in the real world. - A background in embedded software, robotics, or hardware-facing engineering, ideally on a product operated in real-world conditions. - You know how things break in the field (power, connectors, storage, clocks, networks) and you design so they don't. - Comfortable on the bench with your tools, and in the field with a laptop and a terminal. - A track record of end-to-end ownership: you've run something in operations, been called when it failed, and made it better. - Familiar with recent LLM techniques (agents, tool use, coding assistants) and how to apply them to real workflows. - An engineering degree or equivalent practical experience. We care about what you've built, not where you studied. - Bonus: you've worked with a fleet of connected devices. **Show us what you built** We hire people who build things nobody asked them to. A company you started, a side project that runs, a rig on your desk, open source you contribute to, a product you shipped in a previous job and kept alive. Send us the link. It matters more than the line on your resume. **Soft Skills** - **Fast learner.** You don't need to have done exactly this before. You need to pick up whatever the problem calls for, quickly. - **Reliability-obsessed.** You'd rather build a check than hope it works, and you write the runbook so the next person doesn't need you. - **Operator empathy.** You care how it feels to use your software in a cold field at six in the morning. - **Autonomous.** Hand you a fuzzy problem and you come back with a v1 and the right follow-up questions. - **Field-ready.** Comfortable in a fast-paced, multidisciplinary team, and happy to go to the trucks when that's where the problem is. - **Languages:** English is required. French is a plus. ## AI Leverage We're building HyLight as an AI-native company. That's not a buzzword: it's how we work. - You use AI tools fluently every day: Claude Code, Cursor, Copilot, or whatever gives you the edge. - You look for places where AI can replace a workflow entirely, like log triage, field diagnostics, or test generation. ## What We Offer - Potential to become a shareholder in HyLight. - Real ownership and direct impact from day one. - An excellent work environment: an international team and beautiful offices on a former airbase, with flights happening right outside the window. - Bi-annual offsite trips with the whole team. - 100% reimbursement of your public transport pass (Navigo), or €800/year for sustainable commuting. - A bike provided for commuting. - Free lunch. - Access to a sauna and gym on-site. ## Why You Should Join - **Founder mentality:** we value extreme ownership and want people who think like founders: fast decision makers, natural problem solvers, and executors. - **Ground floor opportunity:** you'll help shape and grow a revolutionary company with the resources to back it. - **Impact and speed:** your software runs on every airship we fly, and the data it captures is the start of everything our customers see. ## Interview Process 1. Introductory phone call (20–30 min) 2. Technical case study, followed by a debrief with the Head of Data 3. On-site meeting with the team (3–4h) 4. Reference calls ## Location - Office: Base 217, Brétigny-sur-Orge, accessible via RER C. Remote work is part of how we operate, but regular presence on the base and occasional field trips with the trucks are part of the job. - Lab and warehouse access. Visa sponsorship is not available for this role.