AI engineer with six years building and shipping production ML and AI systems — for clients in retail, energy, the public sector, and defence.
I designed and deployed an LLM text-classification system that runs 2000+ records a day in production connected to BigData flow, and co-developed an MLOps platform that the client’s Data-Science department adopted as its standard for building and deploying models. I currently lead the architecture of an agentic system with a broad range of tools for qualitative and quantitative data analysis.
I move between implementation — Python, GCP, Vertex AI — and the architect’s role, and I translate technical decisions for people needing the created systems.
This site collects that work, the stack behind it, and a running knowledge base I collect improving my craft.
A Gemini-based text-classification platform scaled from an 11-category PoC to 61 production categories, processing 1,200–3,200 records a day.
Co-developed an internal MLOps platform adopted by data-science department as its standard for building and deploying models.
Operational-level multi-agent battlefield simulation; profiling and optimization cut resource use ~5× and made it ~5× faster.
Key technical concepts I collect and refine as part of a longer effort to get deliberately better at the craft. Pick a topic, then a subtopic, and expand the ones you want.
Language internals and core semantics, the data model and class machinery, concurrency, testing, and idiomatic design.
22 subtopics · 273 concepts
Transformer internals, attention, training and fine-tuning, sampling, inference serving and quantization, prompting and context.
17 subtopics · 171 concepts
Tool calling and the agent loop, MCP, skills and harnesses, multi-agent architectures, and how to evaluate agents.
4 subtopics · 40 concepts
Lexical and semantic retrieval — BM25, embeddings, vector indexes, distance metrics, Elasticsearch, hybrid search, and the RAG pipeline around them.
11 subtopics · 81 concepts
Rate limiting, caching, queues, consistency, database scaling, and reliability — the recurring building blocks behind any distributed system.
6 subtopics · 41 concepts
Experiment tracking, model registries, deployment strategies, monitoring and drift detection, and feature stores — running ML as a reliable production system.
4 subtopics · 24 concepts
JVM internals, the object/memory model, concurrency, and classic design patterns — and how the runtime and idioms differ from Python's.
5 subtopics · 24 concepts
The IoC container and bean lifecycle, proxies and AOP, @Transactional, the JPA persistence context, and the database isolation anomalies underneath it all.
7 subtopics · 95 concepts
The platform basics every GCP engineer must know cold — projects, regions and zones, billing — plus the services for deploying ML/AI workloads and the generative-AI and agent stack.
3 subtopics · 20 concepts
How Git really works under the hood — objects, hashes, refs, the index — and the commands that move them: fetch, pull, reset, switch, rebase, cherry-pick — plus branching workflows and CI/CD.
5 subtopics · 38 concepts
How AI assistants and agents help write code and build systems across the software lifecycle — and the controls that keep the output trustworthy.
7 subtopics · 59 concepts