available · remote · full-time or contract · karachi (utc+5)

Backends and RAG systems, built to be fast, reliable, and hard to break.

Senior backend engineer with 5+ years building multi-tenant B2B SaaS, and most recently enterprise RAG: hybrid retrieval, citation-grounded answers, streamed in real time. I own the async pipelines, realtime delivery and tenant isolation that hold up when traffic climbs and the SLA doesn't move.

Python · Django · FastAPI · Celery · Redis · PostgreSQL/pgvector · RAG · AWS · Azure

request lifecyclelive
client api cache queue db
91
API endpoints, RAG platform
1,800+
automated tests
36
async tasks & periodic jobs
5yr
shipping to prod
/work

Selected work

3 roles · 2020 → now
Diya Interactive Karachi, PK Jun 2023 – Present
Senior Software Engineer & Team Lead (Jul 2024 –)  ·  Software Engineer
  • Team & tech lead for an enterprise RAG knowledge platform (industrial / energy sector), leading 4 engineers (2 backend, 1 frontend, 1 mobile). Owned the architecture and backend: agentic hybrid retrieval with page-level citations, streamed over SSE. → case study
  • Shipped the RAG platform's 91 API endpoints and ~680 tests, a retrieval eval harness with a CI recall gate, and 23-permission tenant isolation. Ingested a 1,300+ page manual into 3,800+ searchable chunks.
  • Team lead and co-tech lead for a security workforce management SaaS, leading 8 (3 backend, 3 frontend, 1 mobile, 1 QA). Core backend across 11 Django apps, 31 ViewSets and 1,200+ tests. → case study
  • Resolved a production memory leak by replacing in-process APScheduler with Celery Beat: 7 jobs migrated, now 12 periodic jobs and 24 task types.
  • Cut repeat bulk-schedule generation from ~9s → 0.06s (thousands of queries down to 16), and moved the schedule table to yearly MySQL range partitions.
  • Redesigned roster distance lookups to replace per-officer Google Distance Matrix calls with a precomputed distance table: 0 Maps calls, an estimated $1.25k–2.6k/month avoided.
  • Built realtime delivery with Redis-backed SSE (Last-Event-ID replay) and Firebase push across 14 event types, and owned the production AWS Elastic Beanstalk infrastructure.
  • Rebuilt a counselling platform's chat as a versioned REST API, upgraded it from Django 2.2 to 4.2, and built its CI test pipeline. Also built a multimodal AI commentary POC (FastAPI, Gemini, FFmpeg).
KNYSYS, LLC Karachi, PK Mar 2021 – May 2023
Associate Software Developer  ·  ex-Junior Software Developer
  • Built and maintained Python/Django backends for distributed business workflows, API-driven apps, scheduled tasks, and automation.
  • Designed REST APIs, database integrations, and Celery/Redis async workflows to improve throughput, background execution, and fault tolerance.
  • Customized and extended large-scale platforms including Open edX: backend workflows, platform config, and integrations across AWS and Azure environments.
Bitrupt Karachi, PK Feb 2020 – May 2020
Full Stack Developer Intern
  • Built React and React Native features and integrated backend services with Firebase and SQLite in an agile team.
/systems

Systems I've shipped

6 case studies · problem → approach → outcome
/fixes

AI code, made production-ready

found · fixed · proven

Code generated by AI and shipped without verification. Here's what was actually broken, and what it took to make it fast and correct.

/stack

What I own

capabilities · toolkit · education
01

AI & RAG systems

Hybrid retrieval (pgvector + full-text, RRF), agentic retrieval, query rewriting, citation grounding, Docling/OCR ingestion and retrieval evals.

seen in → RAG platform
02

API architecture & auth

Multi-tenant REST with DRF and FastAPI, tenant isolation, JWT/OAuth and Microsoft Entra ID SSO, versioned APIs.

seen in → RAG platform · workforce SaaS
03

Async & scheduling

Celery, Celery Beat and Redis for polling, bulk generation and long-running jobs, kept off the request path.

seen in → workforce SaaS
04

Real-time delivery

SSE with Last-Event-ID replay, Redis Pub/Sub and Firebase push, with deduplicated fan-out.

seen in → workforce SaaS
05

Performance & cost

Query profiling, prefetching, bulk writes, MySQL partitioning, caching, and cutting paid third-party API calls.

seen in → 41× schedule fix · workforce SaaS
06

Cloud & reliability

AWS Elastic Beanstalk, Azure VMs and Container Registry, Docker, GitHub Actions, CloudWatch, test suites and production incident response.

seen in → workforce SaaS · RAG platform

Toolkit

LanguagesPython · JavaScript
BackendDjango · DRF · FastAPI · Flask · ASP.NET Core · EF Core · REST · WebSockets · SSE
AI / LLMRAG · hybrid search (vector + full-text, RRF) · agentic retrieval · query rewriting · contextual chunking · citation grounding · embeddings · HNSW · Docling / OCR · retrieval evaluation · Azure OpenAI · Microsoft Foundry
Async / realtimeCelery · Celery Beat · Redis · Redis Pub/Sub · SSE · Firebase Cloud Messaging
DatabasesPostgreSQL · pgvector · MySQL (range partitioning) · MongoDB · DynamoDB · Firebase RTDB
CloudAWS (EC2 · S3 · RDS · Lambda · Elastic Beanstalk · CloudFront · CloudWatch) · Azure (VMs · Container Registry) · GCP · DigitalOcean
DevOpsDocker · Docker Compose · GitHub Actions · CI/CD · nginx · Gunicorn · Linux
AuthJWT · OAuth · Microsoft Entra ID SSO
Testing & toolspytest · Django TestCase · xUnit · Git · GitHub · GitLab · Bitbucket · Azure DevOps
ArchitectureMulti-tenant SaaS · distributed systems · SOA · API design
FrontendReact · React Native · Vite · HTML · CSS
Education
B.S. Computer ScienceDHA Suffa University
Dec 2016 – Feb 2021 · CGPA 3.5
open to senior backend & AI engineering roles

Let's talk about your backend and AI systems.

Remote, full-time or contract. I can join as a full-time hire through an employer of record or work as an independent contractor. Based in Karachi (UTC+5), with full UK and EU overlap and partial US overlap.

Prefer to talk? → Book a 20-min intro call

or reach me directly