Eslam Hamed

Freelance AI Engineer & Solutions Architect

Contract engagement with a Saudi digital-services company — sole engineer on an executive-facing AI analytics product.

• Architected and solo-built an AI data analyst that executives can trust with real numbers: ask a business question in plain language, get an answer where every figure carries a click-to-verify receipt.
• Designed the guarantee behind it — the model never calculates. It decides what to ask, a deterministic engine computes the result, and any claim missing a valid receipt is rejected. With no raw-SQL escape hatch, a fabricated number cannot be produced in the first place.
• Owned the entire engineering function: architecture, build, test, deploy, and the product roadmap the client planned from. Python/FastAPI engine, React/TypeScript UI, PostgreSQL 16 row-level security, an MCP tool gateway (IBM ContextForge), and LangGraph routing every turn.
• Held to team discipline throughout — 387 pull requests, each reviewed and CI-gated before merge, 269 test files running against a real PostgreSQL, and 94 recorded architecture decisions.
• Built Arabic-native analytics: Hijri calendar, Arabic business vocabulary, and a detector for a failure most systems miss — when a business term has two legitimate readings, it answers the likely one and offers the alternative with its own number rather than quietly picking one.
• Built Data Invoker, a synthetic-data engine that derives behaviour from persona and ships a ground-truth answer key with planted signals and deliberate decoys — turning demo data into a benchmark that grades whether an AI analyst found the real insight and resisted the tempting false one.
• Open-sourced the AI-assisted engineering methodology the project runs on as agentic-methodology: a human owns direction, two independent models argue the judgment calls, and tests and CI own correctness — on the explicit principle that model-vs-model agreement is necessary but never sufficient.

Technical Lead & AI Solutions Architect

Enterprise AI platforms. Formal tech lead 2021–23; architect and hands-on AI engineer throughout.

• Managed a 5-person cross-functional team (backend, frontend, QA, product owner) from 2021 to 2023 as formal tech lead — performance reviews, hiring, 1:1s, mentorship — then moved to a broader hands-on architect scope.
• Owned production release engineering for a multi-service platform: releases ship in gated waves where UAT runs the identical image artifacts, verified by digest rather than rebuilt lookalikes, and nothing promotes until post-deploy assertions pass and a soak holds against baseline error rate. Waves ship rollback-free, including unattended overnight runs.
• Brought 35 Azure DevOps projects and ~224 repositories under one governance standard — audited branch policies estate-wide, then executed the unification: mandatory review gates on tier-one repos, protected tags, a hardened deploy-render gate, break-glass access, and a scheduled drift audit. Cleared every blocking finding in an independent security review.
• Ran the model economics for the agent platform — found the orchestrator model was 75% of workflow cost, then proved the cheaper option wasn't the worse one with a frozen-suite comparison at 95% confidence: 73% lower cost per conversation with no measurable quality loss, rolled out across 840 agents with a guarded apply and one-command rollback.
• Architected and solo-built the MVP of Auto-RAG, an automated retrieval-augmented platform that shipped to production for 7–10 enterprise customers including Gulf-region government ministries; later contributed to the production Python rebuild — LangChain/LlamaIndex orchestration, Weaviate and Azure AI Search retrieval, and automated experimentation across retrieval strategies and prompt templates.
• Built MerlinX, DXWand's flagship agentic conversational AI SaaS, in production with ~10 active customers — an agentic framework with tool calling and multi-step reasoning on top of Auto-RAG. Owned full-stack delivery across .NET backend, Svelte/TypeScript frontend, Python AI layer, and Meta channel integrations (WhatsApp, Instagram, Facebook).
• Built an AI-powered RFP analysis system that extracts scoring models from RFPs and evaluates vendor proposals against them, replacing manual review with 80% better comparison accuracy, validated by the client in production — with guardrails, hallucination mitigation, data privacy controls, and human-in-the-loop review.
• Led delivery of Cogent, an Arabic-dialect NLP chatbot SaaS handling 100K+ monthly sessions at 91% intent-classification accuracy across multiple enterprise deployments. Product since retired as the company moved to LLM-based architectures.

Senior Software Engineer | AI Platforms Engineer

Government-scale AI delivery across speech, search, and retrieval systems.

• Delivered the real-time speech-to-text portal with speaker identification for the Egyptian Ministry of Justice court system — selected by the Minister over competing vendors and deployed to several dozen courtrooms in production.
• Delivered the semantic legislation search platform for the Egyptian Cabinet IDSC as prime sole vendor — owned 75%+ of the technical work plus the full service-delivery role from pre-sale through handoff, with the codebase sold to the client.
• Built Preneurs, an LLM-powered pitch-deck analyzer that reduced investor matching time by 60%.
• Built automated evaluation workflows to benchmark retrieval configurations against curated query datasets.

Software Engineer

Backend development and infrastructure at one of MENA's leading health tech platforms.

• Integrated Fawry, Payfort, Salesforce, and Cequence to support financial and partner operations.
• Developed referral workflows, event logging, and user interaction tracking for growth initiatives.
• Built a Redis-powered feature flag tool to accelerate A/B testing and rollout cycles.
• Managed messaging infrastructure using RabbitMQ and AWS SQS while supporting Lambda, CloudWatch, and SNS.
• Supported migration to .NET Core and introduced GraphQL for more flexible API development.

Software Developer

Software delivery for Egyptian Exchange trading systems.

• Developed the online trading dashboard for the Egyptian Exchange (EGX), Egypt's national stock exchange — live market data, order execution, and portfolio insights, while improving UX and memory performance.

Founder & Solo Engineer

Solo build, 14 calendar days, live product with paid tiers — careerrafeeq.com

• Shipped an end-to-end production AI SaaS solo in 14 calendar days — backend, frontend, Chrome extension, payments, and deploy infrastructure. ~75K lines of hand-authored production code, with 53 integration test files gating every deploy.
• Designed an 8-phase deterministic LLM orchestration pipeline with parallel fan-out and a confidence-gated refinement loop, plus a multi-run statistical eval harness gating every prompt change.

Computer Science Bachelor

Very Good with Honors. Graduation project: a Shiny application supporting data cleansing and visualization workflows for the data science process.