Notes, experiments, and playbooks on building production LLM applications, orchestrating autonomous agents, and optimizing software delivery. Written for developers and practitioners.
Learn how to architect, build, and deploy production-ready AI agents using the Model Context Protocol (MCP). Covers state management, security, and scalability.
Learn how to proxy AWS Bedrock API calls to local open-source models using MiniStack and Ollama, enabling zero-cost, offline generative AI testing.
A definitive engineering guide to building, evaluating, and deploying production-grade AI agents using the NVIDIA NeMo Agent Toolkit and NVIDIA NIM.
An analysis of the state of enterprise AI agents. Covers the shift from single-agent to multi-agent architectures, the rise of MCP, and edge inference.
A definitive guide to defining, orchestrating, and executing AI workflows. Covers sequential chains, parallel execution, and human-in-the-loop architectures.
AI Engineer & Systems Practitioner
Building LLM products, researching multi-agent orchestration, and setting up secure deployment pipelines. Formerly writing DevOps playbooks, now mapping the production AI stack.
Learn more about my experiments arrow_forwardEverything published here is tested locally and run against real applications. We avoid corporate jargon, consultant hand-waving, and empty marketing hype.
Designing, optimizing, and operating reliable AI pipelines in production. Context truncation, pricing model controls, hybrid search architectures.
Orchestrating autonomous decision-making loops and workflows. Complex state machines, multi-agent frameworks, LangGraph orchestration.
Practical implementations, fine-tuning methodologies, model evaluation frameworks, and speech-to-text models on Azure and cloud environments.
Accelerating local development velocity, command-line interfaces, automated CI/CD pipelines, container optimizations, and secret scanners.
We send out weekly breakdowns of RAG implementations, agent logic flows, Docker optimizations, and code checklists. Straight code snippets, zero fluff.