Patryk Ruciński Available for contract work

I build AI systems that have to be right the first time.

Founder and lead engineer of INTELIT.AI, where twelve AI agents read freight documents, plan loads and verify carriers against government registers for more than twenty transport companies. In that industry a wrong answer is a legal liability, not a demo bug — so the work is evaluation, validation and exception handling as much as it is models.

Warsaw, Poland. I work US East hours or CET, on deliverables rather than tracked time.

Illustration: a carrier's consignment note is scanned, five fields are extracted with confidence scores, four are verified against the national court register, the EU transport licence register and the insurer's API, and one — the insurance coverage limit — is flagged for review because it is below the value of the load.

  1. Carrier0.99 TransKol Sp. z o.o.verified national court register
  2. Licence no.0.98 PL-0421/2019/CLverified EU transport licence register
  3. Insurance policy0.97 OC-P 55 118 903verified insurer API
  4. Premium paid0.96 12.01.2025verified insurer API
  5. Coverage limit0.94 €100,000flagged for review insurer API below load value €142,000

4 verified · 1 flagged

What I do

Selected work

INTELIT.AI — AI platform for freight

2024–present · founder, architect, sole builder

A multi-tenant transport management system on AWS with twelve production AI agents: order intake from email and freight exchanges, load planning, ETA and delay detection, invoicing, driver settlements, and document verification against government and commercial registers. I designed it, built it, and I run it — including integrations with TimoCom, Trans.eu, telematics providers, Polish e-invoicing and Comarch ERP.

  • 20+transport companies running on it
  • Thousandsof orders and invoices a month
  • Up to 80%less time spent processing an order
  • 3–5 daysof additional cash flow from automated invoicing
How the carrier document verification works

Before a European freight forwarder can onboard or pay a carrier, someone verifies a stack of documents by hand: cargo insurance and its coverage limits, the EU transport licence, proof that the premium was actually paid, posting declarations for cabotage and cross-trade. In practice that means reading dates and limits off photographs taken by drivers and cross-checking them across several systems.

The system extracts structured data from those documents, checks coverage against the load, confirms licence validity and premium payment, and then verifies the result against public registers and regulatory APIs rather than against what the carrier claims. Only failed checks and low-confidence cases reach a person, and every decision is explainable and auditable.

Onboarding that took half a day now completes in minutes. The parts that mattered were unglamorous: validation against real registers, confidence thresholds tuned to how operators actually behave, and an exception path people trust enough to stop double-checking.

Frozex — road freight and forwarding

2023–present · owner

I own and run a freight business: a fleet of vans and rigid trucks plus a subcontractor network, operating across Poland and the EU. It is why the edge cases in INTELIT were in the design from day one instead of being discovered in production — I automate problems that are on my own desk.

Background

  1. 2024 – nowINTELIT.AI — founder, CEO & lead engineer
  2. 2023 – nowFrozex — owner, road freight & forwarding
  3. 2023 – 2025HERE Technologies — tech lead
  4. 2019 – 2023Independent contractor — CodiLime, CloudBest, ING Bank, ComCERT, Usecrypt
  5. 2016 – 2019Freelance developer — web and backend for small businesses
  6. EducationMSc Data Science, Warsaw University of Technology · BSc Cryptology & Cybersecurity, WAT