It had been a while since I personally shipped a production fix.
Years of product strategy and project management pull you further from the codebase, closer to roadmaps and stakeholders. That’s the job — but somewhere along the way, the muscle of actually building gets quieter.
This week I got it back, for a few hours.
Our WhatsApp OTP service went down in production after a provider migration (AiSensy → MSG91). Real users, real login flow, real “why isn’t this working” pressure.
Using Claude Code, I:
- Traced the failure through logs on the actual production server
- Found the root cause (the number had moved providers, but the code hadn’t)
- Designed and shipped a provider-switch so we could run either service
- Validated it locally before it ever touched production
- Deployed the fix live — with backups, rollback points, and zero downtime
- Tracked the whole thing properly in Jira, end to end
None of it was magic. It was engineering — investigation, tradeoffs, verification, documentation. Claude Code was the pair programmer that let me move at the speed my judgment could keep up with, even with rusty hands.
I’m re-sharpening the technical edge — Claude, AWS, hands-on architecture — alongside the product and strategy instincts I’ve built over the years. That combination feels like exactly where the industry is heading: leaders who can still open the terminal.
More builds coming. 🛠️
Linkedin post: https://www.linkedin.com/posts/vikasjainlive_aiproductmanagement-aiengineering-aiconsulting-activity-7490628101962354689-0DBU
