Jahanzaib Imran
Full-stack engineer building AI-powered SaaS products.
New to Contra
Jahanzaib is ready for their next project!
A scraper returning HTTP 200 tells you the browser worked. It tells you nothing about whether you got the data.
On a property-listing pipeline running 800–1,000 records per run, a single selector change dropped price extraction from a 97–99% baseline to 10–15% — while every request still came back successful. Silent failure.
I build data-contract monitoring into scraping systems for exactly this: expected record counts, missing-field thresholds, and freshness checks that fail loudly when extraction degrades. Plus per-session proxy configuration and browser fingerprint management across 20–30 concurrent sessions, with proxy failures categorised separately from extraction failures so you know which problem you actually have.
Stack: Playwright · Puppeteer · Node.js · Redis/BullMQ · proxy-chain
0
4
Legal ops teams run on three systems that don't talk to each other. The gap gets filled by someone copying fields between tabs.
I built the integration layer that syncs data automatically across HubSpot, Litify, and Filevine — event-driven via webhooks, with field mapping, authentication, duplicate prevention, and error handling for the cases where a third-party API returns something unexpected.
The hard part in integration work is never the happy path. It's what happens when one system is down and the other isn't.
Stack: Node.js · Express · PostgreSQL · HubSpot/Litify/Filevine APIs · webhooks
0
9
Healthcare software has a different failure mode: a bug isn't a bad user experience, it's a compliance event.
I built the admin panel for an FDA-cleared medical device administration platform — patient management, device administration, compliance tracking, and reporting — running at 50,000+ concurrent users in production.
The engineering challenge was less about features than about correctness under load: keeping data consistent and auditable while the system scaled.
Stack: React · Material UI · Node.js · Express · PostgreSQL · JWT · REST APIs
0
13
AI Leasing Automation — Intelligent Conversations for Apartment Communities
Leasing teams lose deals to voicemail. Prospects call about a unit, nobody picks up, and they've called the next community before anyone rings back.
I built the platform that answers instead — AI voice agents that handle inbound leasing calls end to end: routing, live transcription, lead qualification, and sentiment analysis, with every conversation scored and surfaced in an analytics dashboard the leasing team actually uses.
What I worked on:
• Real-time call handling and routing via Twilio, including fixing streaming failures that were dropping audio mid-call
• Speech-to-text pipeline — evaluated and integrated AssemblyAI and Deepgram against the legacy streaming setup
• LLM-based call analysis and scoring with the OpenAI API to qualify leads automatically
• Vector retrieval over historical calls so agents have context from prior conversations
• Analytics dashboard in React/Next.js on a NestJS + PostgreSQL backend, with Redis and BullMQ handling background processing
Stack: React · Next.js · NestJS · PostgreSQL · Redis · GraphQL · OpenAI · AssemblyAI · Docker · AWS
0
26