
DRAWITAR
Published Android app for AR-assisted drawing and sketch tracing.
Public Google Play release combining camera-assisted tracing, a sketch library, timelapse recording, AI tools and custom FastAPI backend infrastructure.
SELECTED SOFTWARE PROJECTS
Mobile and desktop applications, backend systems, local AI, RAG, automation and machine learning experiments. Explore the problem, architecture and engineering decisions behind each project.
14 PROJECTS / 4 FEATURED
01 / SELECTED WORK

Published Android app for AR-assisted drawing and sketch tracing.
Public Google Play release combining camera-assisted tracing, a sketch library, timelapse recording, AI tools and custom FastAPI backend infrastructure.

Personal AI memory system powered by local models and semantic search.
Android and desktop interfaces for recording voice notes, retrieving information with RAG and interacting with a local AI assistant. The backend runs on private infrastructure with PostgreSQL, Whisper, BGE-M3 and Qwen.

Local AI opportunity intelligence + personalized ML R&D.
Local-first system combining offer ingestion, structured LLM analysis, human feedback, reusable capability matching and controlled personalized ranking experiments.
Reusable software and AI module architecture.
Module-oriented system for composing reusable capabilities across application workflows and integrations.
02 / MORE ENGINEERING
Local AI voice tutor for mixed Polish-English speaking practice.
Working R&D prototype combining mixed PL/EN speech recognition, deterministic transcription selection, separate Teacher and Conversation LLM roles, conversation persistence and language-aware Android TTS.
Reproducible preference-learning ML research.
Controlled ML study comparing structured offer features with lexical TF-IDF text features using frozen data, shared folds, leakage-safe evaluation and a formal paired benchmark.
Job offer acquisition and analysis workflow.
Workflow covering job-offer acquisition, cleaning, storage, analysis and ranking or matching.
Automated dubbing R&D pipeline.
Experimental research prototype exploring speech processing, speaker analysis, translation, voice generation and media reconstruction.
Full-stack product and SaaS architecture.
Application architecture for a full-stack product with backend, data, billing and integrations.
Real-time local speech and AI pipeline.
Speech transcription, translation, question routing and local LLM-assisted responses.
IN PROGRESS · AI-assisted mathematics learning platform.
Learning platform direction for structured mathematics practice, guided problem solving and AI-assisted tutoring.
Computer vision board understanding.
Vision pipeline for locating a puzzle board, validating geometry and converting screen pixels into structured state.
Desktop media processing workflow.
Queue-based desktop utility for media inspection, format selection, downloads, retries and output handling.
Multimodal prompt orchestration workspace.
Local tool for composing image and video generation workflows from structured prompts and visual references.