Software architect, 15+ years in production systems, almost entirely startups and internal startups — fintech, e-commerce, pharma, publishing.

The work I get pulled into is the recurring startup problem: a service shipped fast under launch pressure, without adequate tests, that later has to be made reliable without being stopped. Incident response, re-architecture, and the release discipline that keeps it from happening again. Most recently that has meant a regulated UK consumer-credit platform — loan servicing, arrears, forbearance, statutory breathing space, and early-settlement calculations written against consumer-credit legislation. Regulation as code, behind a test suite larger than the production codebase.

I've done that in all three configurations: taking a core system from problem statement to release, leading the team that carried it (1 to 7 engineers in ten months), and now doing the same work again with agentic tooling covering what the team used to.

On the data side: a LightGBM acquisition model over a 38M-row base — 0.77 test AUC, 8x lift in the top 1% — scoring 2.9M households for a live campaign. I also found a validation-set misuse defect in my own pipeline (early stopping on the test split), quantified its effect across every published figure, and added a pure-noise regression test to pin the corrected result to chance. NLP is hands-on rather than API-deep: my degree thesis fine-tuned BERT, RoBERTa and XLNet to state of the art on the FNC-1 stance-detection benchmark, published at LREC 2020.

Building on my own time: github.com/vslovik/fenix — a local market-signal scanner (Ollama embeddings, sqlite-vec, no API keys) that ranks incoming articles against a free-text description of what you're looking for, and answers questions over the same corpus with citations back to the source chunks. Also a tool-calling agent that turns unstructured regulatory text into a deterministic calculation pipeline, where the model does the extraction and a deterministic engine does the arithmetic.

Looking for agentic AI/LLM engineering, LLM evaluation and observability, AI integration, or software architecture. Founding-engineer shape suits me — early enough that I'm in the room where the work gets defined. Employment or named-delivery consulting, not disguised staffing.

15+ years building and running production software, currently a Team Lead / Software Architect at a UK fintech/consumer-services company. The work I actually get pulled into most, beyond the job title: taking over a service after it's already broken and rebuilding it so it stops breaking — production incident response and re-architecture, unofficial but real, no ticket tracker needed, just the git history.

Ten months ago I grew a team from 1 to 7 — hired and onboarded everyone myself, rebuilt the codebase, got real end-to-end acceptance-test coverage running with the whole team's buy-in.

Right now I'm deep in agentic AI/LLM systems as an engineer, not a researcher (there's a PhD in Physics and a published NLP/Transformers paper behind me, but I'm 20 years into building software, not chasing a research title). Currently building: a local-embeddings market-signal tool (Ollama, cosine similarity, zero API cost) that mines sources and ranks them against a role description I wrote myself; a tool-calling agent that turns unstructured regulatory text into a deterministic tax-calculation pipeline.

Looking for agentic AI/LLM engineering, technical architecture, or production-reliability roles — genuine employment or named-delivery consulting.