I’m Oussama Zeddam. I build the software around the AI—and make the whole system work. As a senior full-stack AI engineer working remotely worldwide, I take ownership across architecture, application code, integrations and production delivery. Bring me the product that needs to ship, the platform that needs to evolve, or the AI pilot that needs to become dependable software.
One accountable engineer across the critical path.
Hire a senior full-stack developer for frontend, backend, APIs and AI integration. My work spans SaaS development, AI agent development, RAG systems and scoped fintech software engineering. Explore the Expensify fintech engagement.
Full-stack product ownership
Connect the user experience to the API, database, permissions and operational workflows beneath it. I work across those boundaries so implementation decisions serve the product and the next engineer can understand how the system fits together.
Production AI beyond the demo
Build agents, retrieval and tool integrations around explicit behavior. Ground outputs in approved information, validate consequential actions, and establish the evaluation and observation needed to identify failures before they become routine operations.
Remote delivery with clear ownership
Work directly with the engineer responsible for the decisions and implementation. Expect written scope, agreed milestones, reviewable changes and a usable handover. Communication windows and availability are agreed for your team rather than promised generically.
Strengthening the infrastructure behind production AI agents: fail-fast validation against ambiguous tool dispatch, collision-resistant MCP tool names and application-controlled UI streaming. Implemented, regression-tested and merged into OpenAI’s official Python and JavaScript SDKs.
Viru store · London Book Fair Bookstore of the Year 2022
Catalogue-grounded AI discovery for Estonia’s celebrated bookseller. My work connects reader intent with structured product retrieval and catalogue synchronization, turning open-ended searches into relevant book recommendations.
Full-stack systems engineering for an Estonian EdTech platform with over 5,000 articles and tests in its public catalogue. Coordinated permissions, learning content and operational workflows connect administrators, teachers, students and parents.
Full-stack engineering inside a publicly traded fintech. Root-cause analysis, a focused production fix and regression testing in a large React, React Native and TypeScript application, delivered through a completed five-star Upwork contract.
$2.3B in FY2025 sponsored research support, including SLAC
Selected production software delivery for Stanford University. Structured sentiment analysis, target extraction and evidence-grounded outputs, with abstention rules that keep unsupported inferences out of the decision path. The public technical note explains the architecture.
Nous Research · Repository stars observed November 2026
Merged upstream engineering in Nous Research’s widely adopted agent framework. Deterministic identity routing, alias resolution and per-user cache isolation strengthen multi-user sessions, with setup tooling for operators.
MCP-ready · Public APIs · Client production platform
Architected and delivered a live AI marketplace for a client: searchable skills, prompts and workflows, inspectable provenance, permissions and compatibility, plus public APIs and authentication. A full-stack product built for people and agent clients.
Architected and delivered a client platform that brings multi-model AI agents into business operations. Orchestration, RAG, knowledge connectors, BYOK and guardrails form one system for deploying, supervising and integrating specialized agents.
Explore the project
How we work
Global work. A record that travels with it.
Work directly with the engineer responsible for delivery
Agree on scope, milestones and acceptance criteria
Leave with working software and a usable handover
Professional experience since 2013 includes production software, SaaS, automation and AI systems. My formal foundation is a master’s degree in Software Engineering from Arizona State University. My public record includes merged engineering in OpenAI’s Python and JavaScript Agents SDKs—realtime dispatch, MCP tool identity and UI streaming—and the Exceptional Software Engineer trophy from The Connected Awards, North Africa edition.
I work best when the challenge crosses boundaries: an AI feature that needs the right data, a product that needs coherent permissions, or a release blocked by assumptions spread across several tools. I make the constraint explicit, design the smallest useful architecture and carry the work through implementation.
I am based in Algiers and collaborate internationally. Your project does not need to be in Algeria, North Africa or Africa to be a fit. What matters is the work, the level of responsibility required and whether we can establish a productive operating rhythm.
Before the first conversation
Questions worth asking.
Can I hire you remotely from anywhere in the world?
I work remotely with international companies, including teams in the United States and Europe. Share your location, expected communication windows and project scope so we can confirm availability and a practical working arrangement.
How do I choose the best fullstack engineer in the world for my project?
There is no single universal ranking that determines project fit. Compare engineers against the work: architecture, frontend, backend, integrations, AI behavior and ownership after release. My case studies and public contribution records let you evaluate specific engineering work before we discuss your project.
What is a full-stack AI engineer?
A full-stack AI engineer builds both the AI capability and the application around it: user interfaces, backend services, databases, authentication, integrations and production workflows. My practice covers that end-to-end scope, with responsibilities agreed for each engagement.
What does your OpenAI Agents SDK work demonstrate?
My merged work across the Python and JavaScript SDKs addresses predictable realtime dispatch, collision-resistant MCP tool identity and application-controlled UI streaming. The public pull requests show implementation, regression tests, documentation and upstream review—the same execution discipline I bring to client systems.
Your next production system
Bring the hard problem. Let’s build the way through.
Tell me what needs to work, what is blocking it and what success looks like. I’ll review the fit and help define the next move.