About
Hi, I'm Shahzaib.
The short version
I'm Shahzaib, a full-stack AI engineer from Pakistan. Since graduating from FAST NUCES in 2024 I've been an engineer, often the first or lead one, at AI startups in the US, UK and Europe, across meetings, hospitality, legal tech, AI governance and security.
My best work so far is Bridge. I joined as the first engineer, shipped the MVP, grew it to ~500 users, and now lead engineering across product and growth as Founding Senior Engineer. At DreamDesk AI I lead the team building guest-service agents for Second Nature, DreamDesk's parent company, live across its hometels.
In September 2026 I started Zehn, a side project teaching schools and colleges in Pakistan to build with AI. It keeps me close to the part of a company that isn't code: selling, teaching and earning trust.
Outside of work, I'm driven by my faith. It shapes how I build, lead and treat the people I work with.
How I work
Ship to real users early
Bridge went from first commit to MVP in two months. DreamDesk had a working proof of concept in four weeks. Feedback from people using it beats a perfect spec.
Own it end to end
Architecture, backend, frontend, AI pipelines, infra and CI/CD. I also write the specs, plan the sprints and lead the engineers.
Stack
- AI and agents
- LangChain, LangGraph, LlamaIndex, Pydantic AI, RAG, MCP servers, AWS Bedrock
- Voice
- LiveKit, VAPI, AssemblyAI, Deepgram, TTS and voice cloning
- Backend
- FastAPI, Node.js, gRPC, REST
- Frontend
- Next.js, React, TypeScript
- Data
- PostgreSQL, MongoDB, Redis, Supabase, Pinecone, pgvector
- Infra
- Docker, GCP Cloud Run, AWS EC2 and S3, DigitalOcean, Nginx, GitHub Actions
Education and research
BS Computer Science, FAST NUCES · 2024
Paper · arXiv · Jul 2026
Towards LLM-assisted High-Quality Property Generation for Solidity Smart Contracts
Co-author · with Dr. Hassan (DFKI Bremen) and Dr. Jalil (UAEU)
- An LLM pipeline that generates test invariants for property-based testing of Solidity contracts.
- An agent that automatically fixes compilation errors in the generated properties.
- Evaluated with mutation testing: Gemini 1.5 Pro with prompt chaining reached a 25.99% mutation score vs 31.75% for hand-written properties, and matched them on some contracts.
Questions
What does Shahzaib Khan do?
I'm an AI engineer. I build AI agents that run in production: meeting agents at Bridge, guest-service voice and chat agents at DreamDesk AI, and the RAG pipelines, MCP integrations, backend and infrastructure behind them.
Is Shahzaib Khan open to new work?
Yes. I'm open to roles and contracts building AI products. I work remotely from Pakistan (PKT) and overlap with US and UK working hours. Email hello@shahzaib.ai or book a call at https://calendly.com/shahzaib-khan0217/30min.
What kind of AI systems has he built?
Agentic workflows with LangChain, LangGraph and Pydantic AI; voice agents with LiveKit, VAPI, AssemblyAI and Deepgram; RAG pipelines on Pinecone and pgvector; and MCP integrations with ClickUp, Slack, Google Drive and Notion. I ship them full stack with FastAPI, Node.js and Next.js on GCP, AWS and DigitalOcean.
Where is Shahzaib Khan based?
Pakistan. I work remotely with teams in the US and UK.
What is Bridge?
Bridge (meetonbridge.com) is a workspace where teams and AI agents meet: calls, live notes, memory of people and companies, and agents that join meetings and act on what was said. I joined as its first engineer in August 2025 and became Founding Senior Engineer in October 2026.
Say hello
Want to build something together, talk about AI, or just compare notes?