Hardware Systems Engineer

Visia builds Solstice, which combines cameras, infrared, X-ray and more with compute on site and models we train on each customer's own material. It runs inside steel mills, scrap yards, and recycling plants across the US. As we add cameras and models at each site, we're designing larger edge compute systems that run our vision stack on site, close to the sensors it serves, and we need an engineer to own that hardware. Many of the big engineering calls on the first build are still open, and you'll be the one making them.

This is a hands-on role. You'll do the sizing, the drawings, and the supplier calls yourself, and you'll be on site when the racks go in and the first system is brought up. You won't have reports on day one, but once the builds start to repeat you'll hire and lead the hardware team that builds and commissions them, while still owning the design.

You'll report to James, our CTO, and work closely with our deployments team, who install and service our edge sites, and our platform team, who own the edge software, models, and cloud. Most of your time, around 80%, will go to the new edge compute systems, and the rest to the reliability of the edge hardware we already run at customer sites.

• Design the cooling for each system. Each site has a different power budget, so you'll size the cooling to match and decide how much redundancy is worth the cost.
• Design the power. Work out the path from the site's power handoff to the rack, size it by hand, and check the suppliers' numbers.
• Make it run with nobody there. Make every sensor visible and controllable from our side through PLCs or building controls, and design the system to fail safe and recover fast.
• Build the enclosure. Work with our fabrication partners on the physical specs, so the unit survives a crane lift and holds up next to an operating shredder.
• Build and commission it yourself. Be on site for the install, and run the servers at full load before we hand the unit over.
• Keep the bill of materials up to date, chase lead times, find second sources, and push back when a supplier over-specifies. Bring in licensed engineers for the stamped drawings, and check their work.
• Make our edge hardware more reliable. Find the root cause when the computers or enclosures at our sites fail, and when the same failure shows up twice, change the design for every site instead of patching one.

• You've taken a physical hardware system from design into operation in the field, and you can tell us exactly which parts of it you owned.
• Your mechanical, electrical, or thermal engineering is practical and your hands-on work is recent.
• You can size heat and electrical loads by hand and tell us what your assumptions are.
• You read and draw electrical and mechanical drawings, and you're at home with test equipment, suppliers, and the hardware itself.
• You think in failure modes, design for the day nobody can come out to fix it, and treat cost as a design constraint without giving up on reliability.
• You've picked up a new engineering field quickly because a project needed it.

• You've built or commissioned GPU servers or high-density compute racks.
• You've worked hands-on with liquid cooling.
• You're comfortable specifying and debugging PLCs, building management systems, or industrial sensors.
• You've rigged and moved skids or other heavy equipment.
• You've run edge computers such as NVIDIA Jetson at harsh sites.

• 1.5 months in: You own the open calls on the first build with written numbers behind each one. You've visited the first site and our fabrication partners and been through our edge hardware failure log. The first design is finished, with power and cooling drawings, the controls and alarm list, the layout, and a bill of materials with every long-lead item quoted. The licensed review is under way and the top fix for our edge hardware is rolling out.
• 4 months in: The first unit is built, commissioned, and running our vision models at full load with nobody on site, and the second is priced from the same design for less. Our edge hardware is down less per unit even as the fleet grows, and you've planned the first hires for your team.