Systems I've built.

Four projects — each starting from a real operational problem, each showing a different pattern for putting AI into production.

Document Search

How I made 2,000 documents searchable in under an hour

Most document archives are effectively unsearchable — a mix of scanned PDFs, images, spreadsheets, and text files with no text layer. Searchzilla is a search engine built for that reality, using AI vision and modern retrieval to find anything, across every format.

RAG Computer Vision FTS Search

Read case study →

AI Research Agent

An AI that argues with itself: building a multi-perspective research system

A fact-checking agent that automatically researches any claim from all sides simultaneously — real browser, up to 50 sources, explicit bias classification. Built to separate verified facts from rhetoric, not confirm a single perspective.

AI Agents Browser Automation Research NLP

Read case study →

Data Pipeline

4,900 records, zero manual entry: AI-powered data extraction from field operations

Field operations generate enormous data in formats no one planned for — spreadsheets with merged cells, multi-tab Excel files, scanned forms, embedded images. Here's how an AI pipeline extracted and validated thousands of records automatically.

Data Pipeline LLM Extraction Validation Computer Vision

Read case study →

Conversational Analytics

Asking questions of your files: a natural language interface for business data

Non-technical teams have more data than ever and less ability to query it. This is a conversational AI system that lets business users ask questions of spreadsheets, PDFs, and scanned documents in plain English — and get sourced answers.

RAG NLP Data Analysis Multi-format

Read case study →