ViewR - solo engineer in stealth

I was the only engineer at ViewR for 6 months, January to July 2025, remote, working directly with an IIT Delhi founder. The company was stealth. The product was a desktop surveillance app for medium-size offices and campuses - face recognition for entry tracking, RTSP camera control, real-time alerts on dashboards.

The stack was Electron for the desktop shell (so we could run on Windows machines at customer sites with no install pain), AWS Rekognition for face recognition (we did not build models, we bought capability), RTSP and ONVIF for camera control across vendors, and a Node backend in AWS that held face embeddings and event streams.

I shipped three things in 6 months. The desktop app. The Rekognition integration with our own enrollment and re-identification logic on top. PTZ (pan-tilt-zoom) control for cameras over ONVIF so an operator could chase a face across the camera grid. Plus a dashboard that listed recent entries, ranked unknowns by frequency, and alerted on watchlist matches.

Then I went on the road. Three pilots in person. Masters Union in Gurgaon (a business school, they wanted to track attendance and unauthorized entries). Giva (a jewellery brand, they wanted security at their experience centers). One more I will not name. Each pilot was 2-3 days on site - install the app, mount cameras, enroll faces, train the customer's ops team, debug whatever went wrong on their network.

What I learned doing it solo - the bottleneck is not coding speed. The bottleneck is decision-making, vendor selection, and customer empathy. When you are the only engineer, every choice is permanent, every bug is yours, every demo is your face.

The founder was sharp. He sold deals, he wrote specs, he picked the customers. I built. The split worked because we were both clear on whose call was whose. Engineering choices were mine, customer choices were his.

I left in July to join Binocs. ViewR is still stealth as of writing. The product worked. The market was harder than the product.