Back to Side Projects

Personal / Side Project · May 2026 – Present

Production-shaped platform work using local Kubernetes, secure service exposure, deployment automation, observability, and AI workloads as hands-on infrastructure test cases.

Overview

A personal applied engineering track focused on building platform systems the way I would expect to operate them in production: reproducible infrastructure, clear deployment paths, service boundaries, tracing, metrics, and backend comparisons. The work uses a local macOS, Colima, Lima, and Kubernetes environment as a fast iteration loop for patterns that map cleanly to cloud or bare-metal infrastructure.

Key Highlights

• Architected a local Colima/k3s AI platform with Cilium, Tailscale Operator, Helm/Kustomize, and phase-ordered deployment automation.
• Built a four-backend vector RAG bake-off across Qdrant, Weaviate, pgvector, and Milvus with shared ingestion, chunking, embedding, search, and smoke-test code.
• Developed retrieval and agent layers with FastMCP, LangGraph, LiteLLM, FastAPI, Prometheus metrics, and OTLP tracing.
• Deployed Langfuse plus an LGTM observability stack for traces, evals, dashboards, logs, and metrics.

Technologies

Categories