Someone has to land the plane.
I'm Sanjay — a technical project manager in San Francisco. I take complicated programs, in wireless infrastructure and in software, and get them live on dates that don't move.
Three that were worth the trouble.
A stadium, four carriers, and a date the NFL picked
mmWave 5G DAS at the Superdome for Super Bowl LIX. Every carrier had its own acceptance criteria, venue access came in narrow windows, and the date was never moving. Most of the work was getting tradeoffs decided in one room instead of across three email threads.
Shipping an AI agent that engineers would actually trust
Led delivery of OpsAI, an AI-powered SRE agent at Middleware, coordinating engineering, product, design and customer success. The hard part wasn't the model — it was agreeing what "good enough to page a human" meant, and building the release process around that answer.
Fifteen builds, four states, one calendar
Concurrent wireless deployments running in parallel across four states, each with its own crews, vendors and carrier sign-off. I standardized how QA and handover worked across 50+ venues, so quality stopped depending on which team happened to be on site.
I started in RF engineering, which is a polite way of saying I spent years finding out where other people's plans met the real world.
Testing coverage in stadiums and office towers teaches you something quickly: the radio is rarely the problem. The problem is fifteen people who each have a reasonable, incompatible idea of what happens next. Learning to fix that is how I ended up in project management rather than engineering.
These days I'm a technical project manager at Middleware, a YC-backed observability startup in San Francisco, working with engineering teams on AI-powered infrastructure tooling. Before that I ran wireless deployments for Tier-1 carriers across the country. Different industries, same underlying job.
Outside of work I follow far too much cricket and NBA, carry a camera I don't use enough, and host the occasional meetup around the city. I'd rather learn something from a documentary than a listicle.
Three things I believe about delivery.
Risk is cheapest early
Almost every project I've seen go wrong went wrong quietly, weeks before anyone said so. I'd rather have the uncomfortable conversation in week one than the expensive one in week six.
One timeline, or none
When engineering, the client and leadership each hold a different version of the schedule, you don't have a schedule. My job is making sure there's exactly one, and that everyone can see it.
Nobody should be surprised
Bad news doesn't improve with age. If something has slipped, the people affected hear it from me first, with a plan attached — not from a status report two weeks later.
If you're building something that has to ship on time, I'd like to hear about it.