Available for Hire

Senior Backend Developer building distributed systems at TB scale.

6+ years of .NET/C#. Built a distributed logging platform processing 3TB+ daily from 300+ microservices. Expert in Elasticsearch, PostgreSQL, gRPC, and low-level performance optimization.

View Projects Contact Me

system_status.log

Throughput 3TB / day

Microservices 300+ Active

Media SLA 99%

SIMD Speedup 15x

Network Activity

hub

Distributed Systems

Designing fault-tolerant architectures with gRPC streaming, Redis coordination, and event-driven patterns for TB-scale data processing across 300+ microservices.

speed

Performance

Low-level optimization including SIMD instructions, zero-copy memory pipelines, and ML-driven compression. Achieved 15x speedup via hardware intrinsics in C#.

verified_user

Production SLA

Building reliable media delivery systems with contractual SLA 99%. CDN integration, fallback mechanisms, and the only service with unit tests in CI.

Flagship Project

gRPC Ingest
Load Balancer

Worker Node

REDIS Hot Cache

ELASTIC Long-term

Distributed Logging Platform

Production

Centralized logging system for .NET distributed applications, processing structured logs from 300+ microservice instances. Elasticsearch storage, gRPC streaming ingestion, Redis buffering. Reduced mean time to detect from hours to minutes. Later formalized as M.Sc. thesis.

3TB+

Daily Logs

300+

Services

MTTD↓

Hours → Minutes

C# / .NET gRPC Redis Elasticsearch Kubernetes

expand_more Technical deep dive

Architecture

Log Collection: gRPC ingestion (unary + streaming) with REST API fallback, structured logging format for .NET application integration
Storage & Indexing: Elasticsearch with index rotation, retention management, and query optimization for TB-scale data
Buffering: Redis for reliable log delivery — cannot lose production logs
No Kafka — direct ingestion to Elasticsearch (intentional architectural decision)

Technical Challenges Solved

Scale: Efficient indexing strategies and query optimization for 3TB+ daily volume
Performance: gRPC streaming for high throughput with batching and backpressure handling
Reliability: Idempotent ingestion, retry mechanisms, Redis buffering for zero log loss
Data Model: Unified log schema with field mapping and metadata standardization across 300+ services

Business Value

Centralized logging for the entire ecosystem. Reduced mean time to detect from hours to minutes. Single source of truth for debugging production issues. Master's thesis based on 2 years of real production data — a rare combination of academic rigor and production validation.

View on GitHub arrow_forward

Selected Works

Media Delivery System

image

High-throughput media delivery serving 100–500k images/day with contractual SLA 99%. On-the-fly transformations, CDN integration, and ML-driven compression.

expand_more Details

SLA 99% delivery — contractual requirement with B2B clients, achieved via CDN integration, fallback mechanisms, and monitoring
ML.NET predictive compression — trained model to predict JPEG quality parameter directly, eliminating iterative brute-force approach
Zero-copy processing pipeline — raw pointer passing from ImageMagick directly to MozJPEG, eliminating intermediate memory copies
Testing infrastructure — the only service in the entire company with unit tests in CI pipeline
Stack: libvips, FFmpeg, MozJPEG, ImageMagick, Azure Blob Storage

C# / .NET libvips ML.NET CDN

Microservices Migration

dns

Led monolith-to-microservices migration with key architectural decisions. Image similarity search via PostgreSQL + pgvector with 15x SIMD speedup.

expand_more Details

PostgreSQL + pgvector — chose relational DB with vector extension over dedicated vector DBs (Pinecone, Weaviate). Hybrid relational + vector search in a single database.
dHash perceptual hashing — image fingerprinting with Hamming distance for similarity detection
SIMD optimization — System.Runtime.Intrinsics for Hamming distance calculation. 15x speedup over standard implementation. Rare skill in the .NET ecosystem.
Architecture design — service boundaries, gRPC vs REST communication patterns, deployment model design

C# / .NET pgvector SIMD PostgreSQL

NationsCities

sports_esports

Real-time multiplayer word game with anti-cheat system and consensus-based voting. Built with Blazor Server and SignalR for live state synchronization.

expand_more Details

Anti-cheat system — browser tab visibility detection via JavaScript interop, progressive penalty system for repeat offenders
Consensus-based validation — players vote on answer validity, majority-rule algorithm for dispute resolution
Real-time state sync — SignalR hub design patterns, game state synchronization across 50+ concurrent players
Dark/light theme, mobile-responsive, product-grade UX polish

View on GitHub arrow_forward

Blazor Server SignalR Tailwind CSS

Experience

media-press.tv group

Junior .NET Developer → .NET Developer → B2B Contractor

Oct 2020 — Present TV Metadata Industry

Backend-focused role in a 4–5 person team handling high-volume multimedia assets for B2B clients (marketing agencies, TV stations, publishers). Progressed from Junior to independent B2B contractor.

storage Designed distributed logging processing 3TB+ daily from 300+ services

image Built media delivery with SLA 99% serving 100k images/day

speed Achieved 15x speedup via SIMD instructions in vector similarity search

architecture Led monolith → microservices migration with key architectural decisions

expand_more Full details & achievements

Key Projects

Distributed Logging Platform — centralized log aggregation for all .NET applications. Elasticsearch storage, gRPC streaming ingestion, Redis buffering. Reduced MTTD from hours to minutes. Later formalized as M.Sc. thesis.
Media Delivery System — on-the-fly transformations (libvips, FFmpeg, MozJPEG), ML.NET predictive JPEG compression, zero-copy processing pipeline. Only service with unit tests in CI.
Microservices Migration — PostgreSQL + pgvector for image similarity search (dHash, Hamming distance). Service boundaries design, gRPC vs REST communication patterns.
Redis Notification Framework — event processing with Redis Streams and backpressure handling.
Universal Search Platform — contributed to API and worker services with vector + full-text search using OpenSearch and Kafka (currently in development).

Impact & Ownership

Go-to person for complex technical problems across the organization
Designed CI/CD pipelines in Azure DevOps for multi-service deployments
Introduced unit testing and CI integration as first in the organization
Presented at internal "dev days" — knowledge sharing across teams
International collaboration (team member from Germany)

Freelance Developer

telewizjatychy.pl

Aug 2019 — Oct 2020

WordPress automation with YouTube integration. Evolved from WinForms GUI to fully automated Windows background service. Early adopter of GPT-3.5 for content tagging.

expand_more Technical details

Automated checking for new YouTube videos and posting to WordPress
Integrated GPT-3.5 API for automated tag and category selection — pioneering practical LLM adoption
Full ownership: design, development, deployment
Stack: C#/.NET, Windows Forms, Windows Background Service, WordPress API, YouTube API

View on GitHub arrow_forward

Why I'm Different

// unique_selling_points.json

{

"performance_mindset": "I write SIMD in production when standard libraries aren't fast enough",

"tb_scale": "Built a 3TB/day logging platform from scratch — not just used one",

"early_adopter": "Integrated GPT-3.5 in production before the hype (2019)",

"ownership": "I don't just write code — I own design, deployment, and the SLA"

}

Tech Stack

Languages

C# .NET SQL

Data & Storage

Elasticsearch PostgreSQL Redis Kafka

Communication

gRPC REST SignalR

Infra & DevOps

Kubernetes Docker Azure DevOps GitHub Actions

Specialties

SIMD / Intrinsics Media Processing ML.NET Blazor

Education

M.Sc. Computer Science

Big Data Analysis

2023 — 2025

University of Economics in Katowice

expand_more Thesis details

Thesis: Platform for Logging, Monitoring & Metrics Analysis in Distributed Applications

Thesis describes a real production system processing 3TB+ daily — not an academic POC
System was built 2 years before the thesis and has been running in production since 2023
Focus on architecture, scaling challenges, and data model design for TB-scale log processing
Rare combination: academic rigor + production validation

B.Sc. Computer Science

Databases & Systems Engineering

2018 — 2023

Silesian University of Technology

expand_more Details

Faculty of Automatic Control, Electronics and Computer Science
Took a dean's leave to balance full-time studies with full-time employment at media-press.tv — demonstrating commitment and time management
Thesis: Program module for visualising trajectory of fifths (music + programming intersection)

Music Education

Classical Violin — Professional

music_note

Classically trained professional violinist. Active performer in a band (violin + cajón), choir member, and pianist. The same discipline from music carries over to engineering — precision, attention to detail, and ensemble collaboration.

Side Quests

NationsCities

Real-time multiplayer word game
arrow_forward
WordpressClient

YouTube → WordPress automation + GPT
arrow_forward
Data Intelligence Platform

Document processing + LLM analysis
lock
expand_more What it does
- Multi-format ingestion: PDF parsing (PdfPig + Tabula for tables), Excel processing (OpenXml), HTML scraping (HtmlAgilityPack), reverse-engineered RSS feeds
- Intelligence: OpenAI API for semantic content analysis, FuzzySharp for duplicate detection
- Resilience: Polly retry policies, circuit breakers, exponential backoff for unstable sources
- Modern stack: SurrealDB (multi-model NoSQL), Telegram bot for real-time notifications
- Full ownership: infrastructure, code, deployment — end-to-end system

About

I'm a Senior Backend Developer from Tychy, Poland, with 6+ years building distributed systems in .NET/C#. My biggest achievement is a production logging platform processing 3TB+ daily from 300+ microservices — a system I later formalized as my Master's thesis at the University of Economics in Katowice.

Outside of distributed systems, I'm a classically trained violinist. I play in a band (violin + cajón), sing in a choir, and play piano. I bring the same discipline from music to engineering — where a single wrong note can collapse an entire symphony, a single race condition can bring down a cluster.

I hold both M.Sc. (Big Data Analysis) and B.Sc. (Databases & Systems Engineering) degrees in Computer Science, and I'm always exploring cutting-edge tech — from pgvector to SurrealDB to early GPT integration.

Languages

🇵🇱 Polish — Native 🇬🇧 English — C1 Fluent