The Role of Markdown Files in the Harness
Small markdown files do not change model weights, but when and how the harness injects them decides whether they act as constraints or as noise.
/Cannot know me simply from here/
I'm a researcher specialising in LLM systems. At Huawei's Edinburgh research centre, I turn cutting-edge research into production-grade products, building, evaluating, and deploying LLM applications: retrieval-augmented generation, agent memory, and multi-agent architectures.
My background pairs strong Python and applied ML engineering with domain expertise in financial-services NLP.
Curious about my technical thoughts? Scroll down and read words.
Curious about my technical skills? Scroll down and read codes.
Big Data Technology | Advanced Technology for Financial Computing XJTLU | UoE | Huawei UKRD Agent Memory | Multi-agent Systems | Game Theory Algorithms | Agent-based Simulation
Small markdown files do not change model weights, but when and how the harness injects them decides whether they act as constraints or as noise.
Insight is the interest earned on systematic knowledge; framing fixes the ceiling; pitching debugs your understanding; interest funds the whole loop.
Why skills break when they change machines, and how SkVM borrows the compiler/VM playbook to make them portable, verifiable, and self-optimizing.
Markdown over vector databases, model-driven recall over embeddings, and offline consolidation: the design bets inside Claude Code's memory, and why they hold up.
Narrative audit pipeline โ infer what's likely true from what a story leaves unsaid, and surface the key missing questions.
Turn handwritten photos or plain text into structured Markdown and save to a local knowledge base.
Daily RSS digest classified and summarized with an LLM; static HTML/PDF and optional GitHub Pages deploy.
Turn a Markdown vocabulary list into one long British-English MP3 for immersive, passive language learning.