Rishi Raj Borah’s Post

Some of the best engineering work happens on systems that aren't broken. Our deployment engine at Emergent ran tens of thousands of apps and served us well through our fastest growth. But it was built for a smaller, simpler Emergent. The bottleneck was never that it failed. It was that everything we wanted to build next was too expensive on top of it. So we rebuilt it. Two ideas carried the new design. Every deploy is now a permanent, traceable record instead of something edited in place. And going live is a safe switch: the new version comes up beside the old one, gets health-checked in isolation, and only takes traffic once it's genuinely healthy. A bad deploy can't touch a running app anymore. Here's what that actually means for the people building on us. Deploys they can trust. Updates that ship without downtime. Rollbacks that take moments, not a rebuild. Systemic issues that surface in minutes instead of through a support queue. None of that shows up in a single growth metric. It shows up in something quieter and more durable: users who trust the platform, stay less frustrated, and keep building here month after month. Reliability is a retention strategy that rarely gets credited as one. And it freed the roadmap. Multi-region, multi-cloud, agentic deployment , the things we couldn't touch before, are all getting shipped on the new foundation. The lesson we keep relearning: get the data model right and half your roadmap turns into consequences instead of projects. As the PM responsible for the platform, that's the kind of bet I'll take every time. Abhinav Gupta https://lnkd.in/dveNunPH