Profile
Bilal Sharif
AI and Full Stack Engineer. Python and Go backends, React and Next.js frontends, agentic LLM systems, realtime at scale, and the GPU and cloud infrastructure underneath.
Recruiter Snapshot
Backend heavy AI and full stack engineer
Five years of production systems. I work best where product requirements meet real system pressure: APIs at 10K requests per second, 50K live sockets, agentic LLM workflows with eval harnesses, self hosted GPU inference, and the recovery work when a vendor dies.
Best Fit
Senior full stack or AI engineer, remote
Strongest fit for teams that need one engineer who can own a backend under load, ship the frontend, put LLM agents into production and run the infrastructure. 5 to 6 hours of overlap with the UK and EU, 4 with US East.
Core Stack
Python, Go, Django, FastAPI, Next.js
PostgreSQL, Redis, pgvector, PgBouncer, Celery, Django Channels, LangGraph, CrewAI, Claude and OpenAI APIs, vLLM, PyTorch, Docker, AWS, RunPod, Caddy, Playwright, pytest.
Work Style
Direct ownership
I scope the problem, find the bottleneck, ship the working path, measure the result and leave the codebase easier to operate. I take the on call and the architecture call.
Recent Roles
Production systems, applied AI, open source and recovery work.
DarkBytes, Senior Software Engineer, Sep 2024 to now. Django, DRF and Go backend ownership at 10K rps, p95 900ms to 220ms, WebSocket layer for 50K connections, 300K job pipelines, agentic LLM systems on LangGraph and CrewAI, self hosted FLUX and LTX on RunPod, Caddy load balancing stack, React and Next.js operator consoles, Playwright in CI, mentoring.
VinnyAndCo, Senior Software Engineer, Sep 2023 to Aug 2024. White label multitenant ecommerce platform on Django and React, an internal operations dashboard in React and Next.js, and a hybrid retrieval RAG pipeline over the product catalog cutting manual brand research up to 70 percent.
AurLab, Software Engineer, Sep 2021 to Aug 2023. API query time cut 40 percent, deploys cut from 25 to 7 minutes on GitHub Actions, Pinecone and OpenAI RAG pipeline saving 30 analyst hours a month, browser extension and session automation, Linux deployment automation.
Open source. Cauli, a Rust and PyO3 task queue at 27,060 tasks per second, 32.8 times Celery. fancydocx, a pure Python DOCX to HTML renderer with zero dependencies that replaced a paid .NET component in production.
Education. BS Computer Science, Superior University, Lahore, 2021.