I build AI agents and the full stack products around them.
I'm Saif Rohyal. I build autonomous, multi-agent AI systems with LangGraph, LangChain, OpenAI Agents SDK, CrewAI, AutoGen and MCP, backed by Python and FastAPI. I have built five end-to-end AI projects, from an e-commerce ops agent to a voice booking assistant, and I am looking for a job, remote role or startup where I can keep building AI products end to end.
My engineering approach is grounded in end-to-end ownership: designing the cognitive reasoning loop (supervisor delegation, state memory, guardrails), exposing it through fast, typed Python & FastAPI services, and delivering clean, intuitive web interfaces.
Hover over or tap any skill to reveal its architectural context and implementation details.
Interact with a live simulation of Saif's supervisor multi-agent state machines, demonstrating autonomous task delegation, error recovery, and tool execution via MCP.
Five end-to-end AI applications spanning multi-agent workflows, computer vision, and real-time voice streaming.
Real-time visual recognition engine that detects, classifies and identifies objects, faces and scenes.
Multi-agent ops system where specialised agents monitor listings, tune prices, write SEO copy, and audit inventory.
Autonomous scientific agent that parses papers, forms hypotheses, executes code in a sandbox, and self-corrects.
Full-stack support platform with live RAG copilot drafting responses, ticket memory, and CSAT telemetry.
Conversational voice agent that handles inbound phone calls, checks calendar availability, and books appointments.
Production engineering contributions in autonomous multi-agent pipelines and backend API infrastructure.
Built autonomous multi-agent production pipelines and backend API infrastructure powering AI assistants.
All credentials and awards are verifiable with corresponding institutional issuing registries.
Recognized as a 'Promising Performer' in the prestigious national Pak Angels Generative AI initiative for engineering novel autonomous multi-agent systems and real-world AI applications.
PAGAI-PP-0266Engineering essays dissecting multi-agent coordination, protocol design, and streaming voice systems.
How to decouple complex business processes into specialized sub-agents orchestrated by a central supervisor, preventing conversational drift and ensuring guaranteed task termination.
Why hardcoding API tools inside agent prompts fails at scale, and how the Model Context Protocol provides an isolated, secure, and standardized bridge between agents and local/cloud resources.
Breaking down the latency bottlenecks in conversational voice agents. Implementing chunked acoustic streaming, optimistic intent routing, and natural audio barge-in interruption.
Currently open to Full Stack AI Engineer roles, remote engineering opportunities, and startup advisory.
Whether you have an autonomous multi-agent engineering challenge, need a resilient FastAPI agent backend, or want to discuss full-time opportunities, my inbox is open.