Abdul Rehman M.

Your AI prototype works in a notebook. Getting it to run reliably, securely, and affordably in production is where most AI projects fail. That's the part I build. I'm an AI Infrastructure & DevOps Engineer with 10 years of production experience, including CoreWeave, a GPU cloud built for AI workloads. I deploy, scale, and operate LLM and ML systems on AWS and Kubernetes, so your AI product handles real users without surprise bills or 2 a.m. outages. Most AI developers stop at the demo. I start where they stop. LLM DEPLOYMENT & AI INFRASTRUCTURE ✔ Self-hosted LLM serving with vLLM, Ollama, and TGI on GPU-enabled Kubernetes (EKS) ✔ Production RAG infrastructure: vector databases (pgvector, Pinecone, Qdrant), ingestion pipelines, FastAPI serving layers ✔ AWS Bedrock and SageMaker deployments with secure IAM, VPC isolation, and secrets management ✔ Autoscaling GPU workloads and LLM cost optimization (model routing, caching, Spot instances) MLOps & LLMOps ✔ CI/CD for ML models and AI apps: automated testing, evals, and safe rollouts ✔ Monitoring and observability for AI systems: Prometheus, Grafana, Langfuse, latency and token cost tracking ✔ LLM evaluation, guardrails, and regression testing before every release ✔ Experiment tracking and model versioning with MLflow DEVOPS & CLOUD AWS · Kubernetes (EKS, AKS) · Terraform · GitHub Actions · ArgoCD · Docker · DevSecOps ✔ Cut infrastructure provisioning time by 80% with Terraform automation ✔ Maintained 99.99% uptime using SRE practices (SLOs, SLIs, automated incident response) ✔ Zero-downtime blue/green and canary deployments on Kubernetes CERTIFICATIONS AWS DevOps Engineer – Professional · AWS Solutions Architect · CKA · HashiCorp Terraform Associate WHO I WORK WITH AI startups moving from prototype to production, teams whose LLM costs are climbing, and companies that need their AI systems to be secure, monitored, and scalable. Tell me what you're building and where it's stuck. I'll reply with how I'd take it to production.