ASKING ANG — INDEPENDENT BACKEND ENGINEER · PRIVATE API REVERSE ENGINEERING
I reverse-engineer private APIs and application protocols, rebuild them as stable backend services, then integrate and deploy them into real products.
The work starts before the code. I capture real traffic, inspect every request, and map authentication, signatures and obfuscation until the protocol underneath stops being a mystery.
Then — and only then — I write the backend that makes it boring: stable services, honest docs, monitoring, and deployments that stay up while everyone else is still reading error pages.
CAPTURE → ANALYZE → RECONSTRUCT → SHIP
EXAPI unifies eight independent platforms behind a single private-API gateway — reverse-engineered clients, queued workers, Redis caching and a monitored pipeline, serving live production traffic.
SmuleVault started with a waveform and a locked door. I reconstructed the platform's private media endpoints, built the retrieval backend, and turned raw audio streams into structured, cacheable responses — clips, metadata, and covers served with sub-second latency.
The hard part was never the audio. It was the handshake in front of it.
WAVEFORM → PROTOCOL → JSON → VAULT
Scrapers collect agricultural listings across regions; a normalization pipeline cleans, deduplicates and geocodes them; analytics turn market noise into price curves, supply signals and a queryable dataset.
An automation layer for a messaging platform: incoming messages are parsed at protocol level — framed, signed, decoded — then routed through a rule engine and LLM helpers, and answered in milliseconds.
Full conversations handled end to end, with no human in the loop and no official bot API in sight.
MESSAGE → PROTOCOL → BOT → AUTOMATION
I don't just wire up a model API. I design the backend around it — prompt pipelines, context stores, streaming endpoints, cost and rate controls — then deploy it into existing products, from a single chat widget to a fully automated workflow.
Collect → normalize → store → serve. Malformed pages in, structured records out: resilient scrapers, schema enforcement, incremental updates — and APIs on top of datasets that stay fresh without anyone babysitting them.
The web is an undocumented database. I build the extraction layer it's missing.
WEB → SCRAPER → NORMALIZATION → DATABASE → API
Dashboards, landing pages and admin tools that sit on top of the backends I build — assembled, deployed and kept alive. A supporting craft, done properly.
Unknown request → analyzed → reconstructed. Auth, signatures, obfuscation — mapped, documented, reproducible.
Requests enter, responses leave — for years. Versioning, rate limits, monitoring, uptime worth trusting.
From a prompt in a notebook to a model serving real traffic — inside software people already use.
Raw web data in, clean structured data out — pipelines that keep datasets fresh on schedule.
Interfaces for the systems I build — assembled, deployed, and very much alive.
Every technology orbits a real backend architecture. Hover — or tap — one, and its part of the system lights up.
HOVER OR TAP A TECHNOLOGY — THE RELATED PART OF THE ARCHITECTURE LIGHTS UP
Tell me what you're trying to access, automate, integrate, or build. I'll look at the technical problem first.