Wiktor Wilczyński

I build AI, data and analytics systems - end to end, and into production.

Contact

What I do

My work sits where AI, data and software engineering meet: analytics and machine-learning models, the data pipelines that feed them, the APIs around them, and the interfaces people actually use. I build end to end rather than stopping at a prototype - everything below runs on infrastructure I set up and maintain myself. AI is also a tool in the building itself, not only the subject: design review, code review, and the second pair of eyes a solo developer does not otherwise have. That is much of why one person can keep this many systems running.

Live systems

SporeHost

Booking and availability system In production

A guesthouse taking reservations through several external booking channels needs one place to see availability, and needs those channels kept in step with one another.

A booking and availability management system, in production for Przysiółek Spore. Availability syncs with external booking channels over iCal, so dates taken on one channel are reflected on the others.

Not linked: this is a live production system for a real client and sits behind authentication, so there is nothing an anonymous visitor could usefully see.

FastAPI · React · PostgreSQL · Docker · GCP

Admin timeline view - a Gantt chart of nine cottages across two months, with colour-coded confirmed and pending bookings, a maintenance block, and a reservation imported from the SlowHop channel.
Admin timeline: nine cottages, bookings and blocks on one rolling calendar.
A reservation list with a detail panel open, showing the guest's name and contact details, arrival and departure dates, number of guests, amount due, and controls to change payment and booking status.
Reservation detail: guest, stay, payment status and status controls.
The iCal synchronisation dashboard - each cottage has an outbound calendar feed and an inbound SlowHop link, an active toggle, and a green OK status with the time of the last successful sync. The outbound feed URLs are deliberately blurred.
iCal sync: a two-way calendar link per cottage, with last-sync status.
The public booking form - a two-month calendar with a three-night stay selected, and steppers for the number of adults, older and younger children, and whether a pet is coming.
Guest booking, step one: choosing dates and party size.
The public booking form showing six cottages available for the chosen dates, each with its capacity, features, total price, and an expandable breakdown of the nightly rate and the mandatory breakfast charge.
Guest booking, step two: available cottages with an itemised price.

Retail_DB_Calc

Data analytics and ML teaching app Live

There is a gap between reading about an analytics technique and running it against a database large enough to behave like a real one.

An interactive data-analytics and machine-learning application: ten lessons, a live SQL sandbox over a 200,000-row retail database, RFM segmentation, BG/NBD customer-lifetime-value modelling, and sales forecasting with SARIMA and XGBoost.

FastAPI · React · SQLite · Docker

Open the live appretail.wilczynski-ai.dev

Also built

Smaller tools and services. Source is linked where the repository is public.

Agentic_IG_Tool

Instagram content pipeline

An agent-driven content pipeline for the same guesthouse client: pick a photo, draft a caption, route it through a Telegram review step, then publish. Operational.

FastAPI · Anthropic SDK · Gemini · SQLite · Instagram Graph API

browser-server

Browser automation server

A Playwright browser server exposing a single persistent session to an agent over HTTP, with SSRF guards and authentication.

FastAPI · Playwright · asyncio

SourceGitHub

b2b-prospector

B2B prospect finder

A command-line prospect finder with a pluggable search backend, a cached HTTP layer and a strict test suite.

Python · httpx · typer · pydantic v2 · SQLite · GitHub Actions

SourceGitHub

Technologies

Languages

  • Python
  • SQL
  • JavaScript
  • HTML
  • CSS

Machine learning & data

  • pandas
  • NumPy
  • scikit-learn
  • XGBoost
  • statsmodels / SARIMA
  • BG/NBD lifetime value
  • RFM segmentation

Backend & APIs

  • FastAPI
  • REST APIs
  • iCal integration
  • Alembic migrations

Data stores

  • PostgreSQL
  • SQLite

Frontend

  • React
  • Vite
  • Semantic HTML
  • Responsive CSS

Infrastructure

  • Docker
  • Docker Compose
  • Caddy
  • Google Cloud Compute Engine
  • Linux (Ubuntu)
  • Git