Muntazir F.

I build and ship AI agents and automation pipelines — then make sure they actually run in production: fast, cost-controlled, and able to survive a compliance review. I've done this inside a top-5 Canadian bank and its AI research lab, so I know what "passes security review" actually means. Top Rated Plus · Expert-Vetted · $200K+ earned · Toronto, full US business-hours overlap WHAT I BUILD · AI agents & automation pipelines: graph-RAG and document-extraction agents where every fact traces back to its source passage - schema-validated, not vibes · Self-hosted LLM serving: vLLM and KubeAI on Kubernetes (Llama 3, DeepSeek-R1), continuous batching, autoscaling and scale-from-zero, behind an OpenAI-compatible API so your apps don't care which model is underneath · MLOps platforms & DevOps: MLflow, Argo CD, Terraform and Helm, so teams can onboard new ML workloads in days · Kubernetes and OpenShift platforms: migrations, on-prem clusters, Istio, and production observability with Prometheus and Grafana RECENT WORK · Joinable: self-hosted Llama 3 and DeepSeek on vLLM + KubeAI with an OpenAI-compatible API · Fylo: neurosymbolic knowledge-graph pipeline extracting structured, cited records from scientific PDFs · SkanAI: production ML infrastructure on Kubernetes · S Inc: on-prem OpenShift with Nginx Ingress and Istio · BrandWell: Docker Swarm to Kubernetes migration · Terra Dotta: two years as their Site Reliability Engineer · A top-5 Canadian bank: hybrid AKS/Anthos migration of a critical API platform under formal bank change management · That bank's AI research institute: SLURM on NVIDIA DGX, OpenShift CI/CD for ML pipelines GOOD FIT IF · You need an AI agent or automation workflow built end-to-end and shipped, not just prototyped · You're spending heavily on API tokens and want to run open models on your own GPUs · Your LLM prototype works, but latency, cost, reliability or monitoring aren't production-ready · You're in a regulated industry (banking, healthcare, government) and need infrastructure that passes security review · You need a hands-on engineer on a broader dev team who also owns the cloud/K8s/CI-CD side, not a narrow specialist who only does one layer · You need a senior backend/infra engineer, full stop - AI is a nice-to-have, not a requirement HOW I WORK I usually start with a short paid assessment that ends in an architecture and a fixed plan, yours to keep whether or not we continue. You work with me directly, not a team of subcontractors. Documentation and handover are part of every engagement. Write-ups of the Joinable and Fylo systems are in my portfolio. Stack: vLLM, KubeAI, Triton, ONNX Runtime, MLflow, Kubernetes, OpenShift, Terraform, Argo CD, Helm, Airflow, Prometheus, Grafana, Vault, AWS, GCP, Azure, SLURM Certified: AWS Solutions Architect, HashiCorp Terraform Associate, Red Hat EX180, CCSK · MEng Software Engineering, McMaster University Send me what you're running and where it hurts, and I'll tell you how I'd approach it.