I'm Kamil Marszałek, a Computer Science student at Warsaw University of Technology with a strong interest in backend development, systems programming, and applied AI.

• 🎓 Computer Science student at Warsaw University of Technology
• 💻 Interested in backend engineering, APIs, databases, and systems-oriented software
• 🧠 Also exploring machine learning and optimization through academic and personal projects
• 🚀 Currently looking for opportunities to grow as a Software Engineering Intern

I enjoy building practical software projects in Go, Java, Python, C++, and Rust — from CLI tools and backend systems to algorithmic solvers and data-oriented applications.

A terminal-first ecosystem for Microsoft Teams built around a reusable Go library, a Python wrapper, and a CLI/TUI application.
Worked on backend abstractions, Microsoft Graph API integration, caching, and reference resolution for terminal workflows.

A full-stack platform for car listings, purchases, and real-time auctions built with Go/Gin, PostgreSQL, and Next.js.
My contribution focused on the backend and database layer, including authentication, REST APIs, real-time updates, and database-side optimization.

A web tool that optimizes keyboard layouts for a given text against ergonomic metrics (same-finger bigrams, finger travel distance, home-row usage, hand alternation, row jumping).
The optimization core is written in Rust, compiled to WebAssembly with wasm-pack/wasm-bindgen, and uses simulated annealing; the Vue.js / TypeScript frontend handles layout and heat-map visualization.

A Kotlin desktop app for analyzing and visualizing historical stock data, built with Compose Multiplatform and Ktor.
Downloads OHLCV candles from the Financial Modeling Prep API, draws interactive candlestick charts, and computes SMA, EMA, and RSI indicators locally, with indicators and statistics implemented as pure functions over immutable data.

A 3D Mandelbulb fractal renderer combining a Python CLI (Typer) with a Rust computational core via PyO3 / maturin.
Uses ray marching and Rayon for parallel rendering (~11× speedup over sequential), and outputs either a PNG image or a 60-frame animated GIF.

A collection of six PyTorch machine-learning projects spanning tabular data, computer vision, generative modeling, sequence models, and NLP.
Includes a regression model for rental demand, an Optuna-tuned MLP for imbalanced classification, custom CNNs for 50-class image recognition, a class-conditional diffusion model, a bidirectional GRU for composer classification, and a transformer ensemble for Polish hate-speech detection.

A solver framework for a constrained optimization problem on a fixed-height grid using dynamic programming, A*, and heuristic methods.
Includes reproducible experiments, result aggregation, plots, and automated tests.

Android app for measuring vertical jump parameters using smartphone sensor data (accelerometer + gyroscope). The app detects jump phases, estimates jump height, visualizes acceleration, velocity, and height over time, and stores previous results for later analysis. Writen in Kotlin