Full stack engineer across product, infrastructure and security.Most of what I have built is still running.
Greenfield work is fun, but most of my ten years has been spent inside systems somebody else started. Feeds that arrive malformed, databases that grew without a plan, sites that got compromised over a weekend. I like the constraint of having to keep something running while changing it underneath.
I work across the whole stack because the interesting failures happen between the layers, not inside them.
React, Next.js, design systems, motion.
NestJS, Laravel, Node. Schemas and migrations.
Docker, managed Postgres, CI, deploys.
Incident response and cleanup.
MostofwhatIhavebuiltisstillrunning.Thatisthepartofthejobnobodyphotographs.
Themigrationthatdidnotlosearow.TheintegrationthatsurvivedthevendorchangingtheirAPIwithouttellinganyone.Thesitethatstayeduptheweekitwentviral.
Noneofthatisvisiblefromtheoutside,whichiswhytherestofthispageshowsyouthesystemsratherthanthescreenshots.
01
Full stack and infrastructure
A public tool that audits a website and turns the result into a qualified lead. I built the analysis backend, the reporting frontend, and the deployment path that runs it in production and staging.
The audit runs long, so the API answers immediately and scores in the background. The frontend polls rather than holding a connection open.
02
Backend and platform
An ordering platform rebuilt from the previous year's codebase. Split frontend and backend deployment, managed Postgres, object storage for assets, and an end to end test harness that runs against real user roles.
Product media is served straight from object storage to the browser rather than proxied through the API, which keeps request latency off the catalogue pages.
03
Architecture and integration
A headless rebuild of an agency site fed by a property feed that arrives as XML over FTP. The interesting problem was not the frontend, it was making an unreliable upstream look reliable to the people browsing listings.
The feed is the weak link, so nothing reads from it live. A scheduled worker normalises each drop into Postgres, and the site only ever reads a known-good copy.
Ten years, still running.
Start a conversation→Contact
talha.tech01@gmail.comAvailable for select client work and senior engineering roles. If you have something already running that needs to keep running, that is the conversation I am best at.
