Computer Vision Engineer

Lead Computer Vision Engineer / CTO - Edge AI (Football Tech)We are looking for the person who will actually build our systemFAIR SYSTEMS makes refereeing decisions in amateur football objective. No VAR room, no TV production, no professional club budget. Just multiple cameras around an ordinary pitch, AI running locally, and a verdict on the referee's watch within seconds.A prototype is running. It works. Now it has to become a product that is reliable every Saturday, in the rain, on a hundred pitches at once. We are looking for one person to make that happen: the person all of our core technical knowledge comes from.What we are buildingMultiple synchronised PoE cameras around the pitchLocal inference on NVIDIA Jetson: detection, tracking, pose estimation3D reconstruction from image space to pitch coordinatesA rules engine that derives match situations from that: offside, goal, throw-inA decision within seconds on the referee's app and smartwatchPrivacy by design: raw footage stays on the edge, only events and short clips go to the cloud. No facial recognition.The problems that will land on your deskThis is where we are honest, because this is the job:Offside is a millimetre and millisecond problem. Sub-frame timing, calibration that holds up all day, and a system that knows when it is not sure.Multi-camera synchronisation and calibration on pitches that are not level, where the goalpost sits slightly differently every time, and where nobody shows up with a tape measure.Occlusion. Twenty-two players in similar shirts constantly blocking each other. Holding on to tracking identity is harder here than in most industrial applications.Latency budget on edge hardware. Everything that makes your model better makes it slower. That trade-off is yours to make.From demo to pitch. Power cuts, network drops, backlight, fog, a camera that gets knocked sideways. Remote monitoring, OTA updates, fleet management.Calls that have to be explainable. A referee needs to defend the decision to an angry assistant referee. That puts demands on what the system returns, not just on accuracy.Building a data flywheel that makes the system better every match weekend, without needing an annotation team of thirty people.What your role isYou are our technical core. Concretely:You define the architecture: edge, cloud, and the dividing line between themYou write code yourself, and a lot of it. For now this is not a management role.You make the build-or-buy calls and defend themYou translate "the referee has to trust this" into technical requirementsYou hire the first engineers around you once that is possible, and you decide who they areYou have a seat at the table on the direction of the company, not just on executionWho we are looking forMust have:Demonstrable experience with computer vision in production, not just notebooks and papers, but systems running outside the buildingMulti-view geometry: camera calibration, homography, triangulation, coordinate transformationMulti-object tracking and the mess that comes with it: re-identification, occlusion, ID switchesEdge deployment: NVIDIA Jetson, TensorRT, CUDA, DeepStream or equivalent. You know what quantisation does to your accuracy.The ability to hold an entire system in your head: from sensor to API to appYou are the owner, not the executor. If something does not work, you are the one who goes and finds out why.Strong plus:Sports analytics, broadcast tech, ADAS or autonomous driving, drones or robotics. All domains where these problems have already been solved once.Experience with hardware in the field: mounting, weatherproofing, power, networkingMLOps: annotation pipelines, retraining, model versioningYou have been the first or second technical person at an early-stage company beforeAn affinity with football (not required, but fun)What you do not need:A PhD. Where you learned it does not matter to us.Our exact stack. If you have built systems like this, you will pick up our tools in a week.What we offerA market-rate salary and a substantial equity stake. You build something of your own without burning through two years of savings.Full technical autonomy. You decide the stack, the architecture, the approach.Remote within Europe, with one caveat: a system that stands on a football pitch is a system you occasionally have to see standing on that pitch. Expect a handful of days per quarter in the Netherlands, certainly in the beginning.A problem that is genuinely hard, in a market almost nobody is looking at. The big players build for the top of the pyramid. The bottom, where most matches are played and most arguments start, is empty.What we are honest aboutThere is a working prototype and there are pilots, we are building rapidly with a roadmap to 2030. In the first months we experiment a lot on multiple football clubs where amongst some of the largest clubs in the world. There is no engineering team around you yet. You build it.If you are looking for a well-defined role on an existing foundation, you will be unhappy here.If you have been thinking for a while "I would want to build this myself, but with someone else handling the market and the business", let's talk.The processIntro call (45 min) with Rutger. Where we stand, what you are looking for, whether there is a click.Technical deep dive (90 min) with an external CV expert present. No puzzles or whiteboard algorithms: we walk through our actual system and your previous work together.A day working together (paid). We take a real problem from our backlog.Terms and role. Salary, equity, expectations. An open conversation.Reply via LinkedIn or email rutger.vanbeek@fair-ref.com One paragraph is enough: where have you built something comparable, and which problem in the list above do you find the most interesting?