🎓 Senior AI Engineer
📍 Netherlands
🔹 Production LLM Systems · RAG & Agents · Decision Intelligence

I design and ship production LLM systems — multi-agent architectures, retrieval-augmented generation pipelines, and evaluation infrastructure — on top of a foundation in operations research, optimization, and probabilistic modeling.

My work focuses on:

• Multi-agent LLM systems and orchestration
• Enterprise RAG with grounded, citation-aware answers
• LLM evaluation and AI observability
• Production AI on AWS using Bedrock, Lambda, OpenSearch, FastAPI, and MLflow
• Decision intelligence where mathematical rigor supports real-world AI products

I’m especially interested in building AI systems that are deployable, measurable, and trustworthy — not just demos.

• 🏛️ Tax Authority Enterprise RAG — enterprise RAG architecture and evaluation framework for regulated tax-domain QA, with secure OpenSearch retrieval, Bedrock-based grading, citation verification, and observability.
• 🤖 JobPilot — AI-powered job orchestration platform with Kanban workflows, personalized job search, automation, and production-oriented full-stack architecture. Live at jobpilot.oploy.eu.
• 🤖 job-agent-backend — FastAPI backend for AI-powered job search, CV matching, tool-calling workflows, and MLflow-traced evaluation on AWS Lambda.
• 📈 mlflow-tracking-server — self-hosted MLflow setup for experiments, traces, prompt optimization, and evaluation in AI systems.
• 🧠 job-analytics-frontend — Next.js frontend for job analytics, AI chat, search workflows, and CV matching.
• 📊 job-market-pipeline — Python pipeline for scraping, filtering, normalizing, deduplicating, and exporting job-market data.

• 🌐 Oploy Platform — applied decision-support platform translating optimization models into practical tools for routing, scheduling, and network design.
• 🔬 2sHMMREM — research code for a two-stage Hidden Markov / Relational Event Modeling framework.
• 🛠️ RepairShop-milp — MILP-based implementation for repair shop optimization.
• ⚙️ ANN-SMB-SemiExpSimOpt — simulation-optimization and neural-network-based research code for process systems engineering.

• Enterprise RAG & Agentic Systems — RBAC-safe retrieval, grounding, citation verification, orchestration, and control loops
• LLM Evaluation & AI Observability — evaluation pipelines, MLflow tracing, telemetry, and prompt iteration
• Production AI on Cloud — FastAPI, Docker, AWS, deployable APIs, and model-agnostic system design
• Applied AI Products — intelligent job-search workflows, CV matching, analytics dashboards, and AI-assisted tools
• Decision Intelligence & Operations Research — optimization, simulation, and probabilistic modeling as the mathematical layer beneath the AI systems

LLMs, RAG and Agents

Development, Infrastructure and Cloud

Data Science, Optimization and Analytics

• Enterprise RAG systems — grounded retrieval, citation verification, and evaluation-gated quality
• Multi-agent LLM applications — orchestration, tool-calling, memory, and reliable workflows
• Evaluation pipelines and AI observability — quality metrics, tracing, prompt iteration, and runtime visibility
• Production AI services on AWS — Bedrock, Lambda, OpenSearch, FastAPI, MLflow, and CI/CD-ready systems
• AI-assisted products — job search, CV matching, analytics dashboards, and decision-support tools
• Decision-intelligence systems — routing, scheduling, network design, and simulation-backed applications

The combination of operations research, probabilistic modeling, and production AI engineering is rare. Many AI engineers lack the mathematical depth for decision systems; many quantitative specialists do not build deployable products. I work where those two worlds meet, designing AI systems that are grounded, evaluated, and production-ready.

• 🌐 Oploy
• 💼 LinkedIn
• ✉️ Mohammad@oploy.eu