Scaling an app to 6,000 users on a tight budget
A FastAPI backend with connection-pool and index tuning on PostgreSQL, load-tested with k6 to validate latency and stability, from diagnosis to production in one week.
Systems optimization, from design to runtime.
I write about and work on software performance: profiling, architecture, and engineering decisions that cut latency and cost in production.
Currently
Turning heavy API calls into millisecond-scale responses via multi-tier caching, I/O batching, and async precomputation.
Availability
Unavailable at this time, with no projected reopening date.
Core stack
Java · TypeScript · MySQL · Linux
Selected projects
Cutting seconds of latency with a cache that wasn't actually caching
A caching mechanism existed in the code but was never activated, every request redid setup work that only needed to run once per process.
Race conditions in payments: distributed locks on a serverless architecture
In an environment where every request can land on a different instance, an in-memory lock protects nothing against a duplicate charge.
Perceived load time in a browser-based 3D game
Deferring audio and 3D models to the right moment to make the game feel ready almost instantly.
A cache that only busts itself when something actually changes
A service worker's cache name was a hand-typed number, too easy to forget to bump on every deploy.
Instant navigation without giving up a site that works with zero JS
Intercepting the click, fetching only what's needed, and swapping just what changed, without depending on JavaScript to function.
Latest posts
Stack & tools
Languages
Python
JavaScript
TypeScript
Java
Go
Tools
Runtime
Node.js
Deno
Databases
PostgreSQL
MySQL
Redis
Infrastructure
Docker
Kubernetes
Helm
Linux
Observability & testing
Grafana
k6