SaaS product company
Conversational Product Assistant
Built the company's product site and embedded an AI assistant into the core product, so users get answers in context without leaving the interface.
Retrieval & documents
Getting AI to answer from your own documents — accurately, with sources — instead of from memory.
Most of the AI systems I build start with documents: medical records, claims, internal manuals. The hard part is rarely the model. It's parsing messy files, splitting them sensibly, finding the right passages, and making sure every answer can point back to where it came from.
- What this covers
- Layout-aware parsing and OCR for scanned and structured documents
- Chunking strategies matched to the document, not one size for all
- Hybrid search, reranking and citation-grounded answers
- Retrieval evaluation — measuring whether the right passages come back
Work in this area
Case study · Legal practices — plaintiff, defense & injuryMedical-Legal Intelligence PlatformHelps attorneys make sense of thousands of pages of medical records — without patient details ever reaching an outside AI service, and with the same answer every time the case is run.Case study · Health plans & payersHealthcare Claims IntelligenceTurns piles of messy claim documents into clean, structured summaries, so physicians spend their time deciding instead of reading.Case study · U.S. government transit authoritySecure Enterprise AI AssistantA private, ChatGPT-style assistant that lets thousands of employees ask questions about internal documents — with every conversation logged and nothing leaving the organisation.
Independent clients
Client RAG & Assistant Builds
Ongoing freelance work: retrieval systems, domain chatbots and end-to-end AI implementations, scoped and delivered directly with each client.
Code you can run
- blueprints / chunkingCompare chunking strategies on your own documents, with Recall@k and MRR.
- blueprints / longdocLong-document retrieval with a heading tree, routing and ordered excerpts.
- blueprints / ocrTiered OCR routing that trusts good text layers and escalates hard pages.
- rag-evalRetrieval metrics and a CI gate that catches real regressions.
Blueprints on this
Blueprint
Chunking strategies, compared
How you cut documents into pieces decides what your AI can find. Seven ways to do it, what each costs, and the order I try them in.
Built & tested23 Sept 2026 · 5 min
Blueprint
RAG for long documents: give it a map
Ordinary RAG answers 'find me the paragraph' well and 'what does this 400-page file say overall' badly. Keeping the document's structure, with summaries at each level, lets one system answer both.
Built & tested23 Sept 2026 · 4 min
Blueprint
Open-source OCR: route pages, don't pick a model
Open OCR models now rival paid APIs, but the best one for a scanned form is wasted on a clean PDF. This pipeline sends each page to the cheapest tool that reads it well and escalates only when it can't.
Built & tested23 Sept 2026 · 4 min
Teardown
RAG patterns: which ones are worth it
There are dozens of named RAG techniques and most teams need four or five. What each pattern does, what the evidence says, and the order I'd reach for them.
Built & tested23 Sept 2026 · 5 min
Other areas
Contact
Working on something
like this?
I'm open to AI engineering, architecture and training work. Tell me what you're building and what the constraints are — that's usually enough to start.
Prefer a short form? Send a project brief
- Taking on new projects
- Usually replies within a day