Igor Kadziela — Resume

Summary

I build end-to-end software systems that combine robust backend architecture, modern web interfaces, data processing and AI.

My work sits where backend architecture, data and the interface have to agree. ResearchHub ties AI answers to the sources they came from. The parcel-analysis platform separates missing data from a confirmed absence of a restriction. obiekt.io, still in its foundation stage, is a tool for building managers.

Contact

Email

i.kadziela03@gmail.com

GitHub

github.com/IgorK834

Open to

Internship or Junior Full-Stack Engineer · full-time

Work mode

Remote, hybrid or on-site · open to relocating abroad

Experience

1. IT Intern, IT department · Iturri Poland

   Current

   • Optimise, fix and ship new features for a kiosk IoT device.
   • Review and plan the architecture of the company’s IoT systems.
   • Maintain the Raspberry Pi device that powers the kiosk.

2. Frontend Developer · Kaks s.c. and MKA Sp. z o.o.

   11.2025 – 06.2026

   Two companies with the same owner; one website for each.

   • Built and deployed kaksbb.pl for Kaks s.c.
   • Built and deployed mkabb.pl for MKA Sp. z o.o.

   Sites: kaksbb.pl · mkabb.pl

Skills

Backend

Java · Spring Boot · Python · FastAPI · REST APIs

Frontend

TypeScript · React · Next.js

Data

PostgreSQL · PostGIS · pgvector · SQLAlchemy

AI / data systems

RAG · Embeddings · Provenance · Source-grounded generation · Structured AI workflows

IoT

Raspberry Pi · IoT system architecture · Kiosk devices

Engineering

Docker · Git / GitHub · CI · Testing · Architecture documentation

Projects

1. ResearchHub

   Active development · Public repository · Not deployed

   A shared workspace where a group collects sources, writes a report and asks an AI that stays grounded in those sources. Charts, tables, calculations and generated claims keep provenance back to the source file, fragment, dataset or executed analysis.

   Role: Sole contributor to the repository.

   • Cited answers, or an explicit “no evidence”. Questions are answered from the selected sources with traceable citations. When the sources do not support an answer, the result says so.
   • Immutable source versions. Replacing a file adds a new version. Older bytes stay downloadable and every analysis records the exact versions it used.
   • Computation in a Docker sandbox. Analysis code runs in a pinned, isolated Python image with filesystem, network and resource limits. Results keep their inputs and run history.
   • Realtime editing, enforced boundaries. Co-editing on Yjs and Hocuspocus, saved through authorized Spring persistence. Workspace authorization is enforced by the server, including internal service calls.

   Source: github.com/IgorK834/researchhub

   Technologies: Java 25 · Spring Boot 4.1 · Maven · Flyway · Python 3.13 · FastAPI · React · TypeScript · Webpack · TanStack Query · Tiptap / ProseMirror · Yjs · Hocuspocus · PostgreSQL 17 · pgvector · Azure Blob Storage / Azurite · Docker · Bicep (Azure IaC)

2. System Analizy Uwarunkowań Działki

   Public repository · Not deployed

   A web application that analyses the planning and investment context of a Polish land parcel. It combines GIS data and spatial-planning datasets, shows the result on a map and produces a PDF report with a verifiable audit package.

   Role: Sole contributor to the repository.

   • Missing data is not an all-clear. Every section has its own status, freshness and provenance. “Not checked” is never shown as “no restriction”, and the headline stays partial until everything in scope was checked.
   • Vector tiles straight from PostGIS. Mapbox Vector Tiles generated from PostGIS for the planning layer; WMS previews are proxied with tile caching; coordinate transformations are handled explicitly.
   • Reproducible reports, verifiable audit package. PDF reports built from immutable analysis snapshots, plus an audit ZIP whose SHA-256 manifest a stand-alone script verifies.
   • MPZP extraction research. Deterministic parsing of local-plan documents, with an optional LLM path that is off by default. Model output is only a labelled candidate for manual verification.

   Source: github.com/IgorK834/system_analizy_uwarunkowan_dzialki

   Technologies: Python · FastAPI · SQLAlchemy · GeoAlchemy · Alembic · Next.js 16 · React 19 · TypeScript · MapLibre · PostgreSQL 16 · PostGIS 3.4 · GIS / spatial data processing · Docker Compose

3. obiekt.io

   In development · Private repository

   A planned SaaS tool for building managers: residents report faults by QR code, voice or e-mail, contractors work from a link, and every job ends with a PDF protocol and acceptance. Only the foundation exists so far.

   Role: Sole contributor to the repository.

   • Three ways in, one case. A QR form, a voice assistant and e-mail all open the same ticket. The reporter gets a case number only after it is saved.
   • Contractors without accounts. A work order is a link. Entry codes appear only after an SMS check on the contractor’s phone and stop working when the job is accepted.
   • AI proposes, people decide. AI sorts and translates reports in Polish, English and Ukrainian. Danger alerts never wait for it, and acceptance stays with the manager.
   • A record of every repair. A short completion form becomes a versioned PDF protocol. Costs stay outside the app; the manager gets a summary of work done.

   Technologies: Next.js 16 · React 19 · TypeScript · Java 25 · Spring Boot 4.1 · Spring Modulith · Python · FastAPI · PostgreSQL + pgvector · Row-level security · S3-compatible storage · Keycloak (OIDC) · OpenAPI contracts · Docker Compose · ElevenLabs voice agents · E-mail in and out · SMS · LLM provider (to be chosen) · PDF generation

Education and languages

Education

Computer Science, University of Bielsko-Biała · since 10.2023, final semester

Current status

IT intern at Iturri Poland · final-semester Computer Science student

Target role

Junior Full-Stack Engineer, or an internship