LLMOps / AI Platform Engineer. vLLM, Kubernetes/OpenShift, LLM evaluation, AI red teaming.
I build the production stack around LLMs. At Red Hat's OpenShift CI Platform team, I own CI/CD pipelines that gate product releases across 30+ repositories and run hundreds of times a day. Most days I'm in Python and the OpenShift/Kubernetes ecosystem. The throughline across my career is the same: I land in unfamiliar territory and ship tools to make the work go faster. At Red Hat that has looked like an LLM/RAG pipeline I built for automated CI failure triage (LlamaIndex, ChromaDB, knowledge graphs, HuggingFace-hosted models for local inference) because I was tired of reading logs by hand; it cut triage time from 15-20 minutes to 2-3 minutes per failure (~85%).
More recently I've been spending time on the agent infrastructure side: containerized multi-agent harnesses for CI failure triage, custom convergence protocols for getting multiple models to agree on a diagnosis, and observability work to understand what these systems are actually doing.
AI/ML: LLM/RAG pipelines, MCP (Model Context Protocol), vLLM, evaluation harness, LlamaIndex, ChromaDB, HuggingFace (local inference), Arize Phoenix
Infrastructure: Kubernetes, OpenShift, AWS, GCP, Docker, Podman, Terraform, Ansible
CI/CD: OpenShift CI/Prow, GitHub Actions, Jenkins, Poetry, uv, pytest
Observability: Grafana Loki, Prometheus, Splunk
Languages: Python, Bash, SQL, TypeScript, Go
Security: gitleaks, OAuth 2.1, RBAC, NIST/RMF, TLS/mTLS