Skip to content
TagoMind
Collaborate with me

Work with me

Consulting and hands-on engineering, focused on two areas, both backed by systems running in production.

Legal & Tax AI

Search, understanding, and reasoning over large legal, tax, and administrative corpora, from domain retrieval to agentic extraction.

  • Domain-specific retrieval, Q&A, and chat with citations
  • Large-scale document pipelines and structured extraction
  • Human-in-the-loop review tooling
AI-Tax

I coordinate research and engineering on AI-Tax, a live Polish legal-AI platform. I strengthened how the team does research and ships: AI search, chat with citations, and agentic extraction over 3M+ documents and 300M+ semantic vectors. The work is published at ACL 2026 and AILaw 2026.

Case details

Applied NLP & LLM systems

Retrieval-augmented generation, extraction, classification, and agents on your real data, with evaluation you can trust.

  • RAG and hybrid semantic search over messy real-world text
  • Information extraction and structuring
  • Evaluation harnesses, guardrails, and agent workflows
Flat Explorer

I built and shipped Flat Explorer end to end: an AI flat-hunting product where LLM agents score listings against your criteria, summarize visits, and compare options side by side. It is live today, built on the Claude API with LangGraph, FastAPI, and Next.js.

Case details
Also available

Computer Vision & Generative AI: vision-language and generative pipelines, from documents to game maps

Graph & representation learning: embeddings, recommendation, knowledge graphs

Agentic AI systems: tool-using, multi-step agent workflows with reproducible evaluation

Finance AI: applied finance intelligence, through the research group I founded at Wrocław Tech

Engagements range from a few days of advisory work to multi-month builds. I work independently and alongside in-house teams, owning a problem outright or plugging into an existing stack.

Engagement

How we work

01

Scope

A short, focused conversation to understand the data, the users, and the real constraints.

02

Prototype

A working proof of concept that de-risks the hard parts before you commit to a build.

03

Ship

Production engineering, evaluation, and handover so your team can own it afterwards.

Have a problem worth solving?

Tell me about your data, your users, and your constraints. I'll tell you straight where AI helps and where it doesn't.

30 minutes · tell me about your data and constraints