Now
A snapshot of what I'm currently working on, learning, writing about, and paying attention to.
Inspired by the /now page movement — a simple way to share what matters right now.
Working On
Evaluating and promoting GitLab Duo, GitHub Copilot, Amazon Kiro, and MCP-based workflows across engineering teams in an enterprise banking environment.
Continuing platform modernization — pipeline standardization, GitLab SDLC governance, and DevSecOps tooling across a large multi-team engineering programme.
Designing reusable AI skills for enterprise engineering tasks — code review, impact analysis, story refinement, test generation — using Kiro and MCP patterns.
Building out this site as a connected professional knowledge portal — adding pages, content, and structured data progressively.
Learning
Completed a rigorous 16-week, 110-hour programme at Johns Hopkins University covering Agentic AI systems, LLM orchestration, multi-agent workflows, RAG, MCP, and practical AI-enabled engineering (Jun 2026).
Deepening practical knowledge of stateful graph-based agent workflows — supervisor patterns, human-in-the-loop design, and long-running agent loops.
Working through RAG evaluation, chunking strategies, retrieval quality, and production-grade RAG architectures for enterprise engineering use cases.
Exploring how MCP standardizes tool-use across different AI coding environments — building small MCP servers to connect agents to real engineering data.
Writing About
Practical MCP server and client implementation for AI agents.
Systematic evaluation and testing of LLMs, RAG pipelines, and agents.
Building stateful multi-agent workflows with LangGraph in Python.
Real-world usage of GitLab Duo for code review and CI/CD workflows.
How large engineering teams can adopt Agentic AI without disruption.
AI Tools in Active Use
Spec-driven development and AI skills for repeatable engineering workflows.
AI-assisted merge request review and code suggestions in GitLab.
Inline code completion and agent mode for .NET and Angular development.
AI coding assistant, reasoning, and multi-step engineering workflows.
Agentic coding via VS Code with MCP tool integration.
This page reflects what I'm genuinely focused on right now — not a marketing list. It's updated periodically as priorities shift.
Last updated: June 2026