Machine intelligence that survives production.
I'm Paweł — an AI engineer with 8+ years of shipping systems for AT&T, NBC Universal, Wasabi, and startups across the US and EU. LLMs, RAG, computer vision — built in Python and Go, deployed on Kubernetes, kept alive in the real world.
The model is the easy part. Everything around it is the job.
Most of my career has been spent in the gap between a promising demo and a system someone actually depends on — data pipelines that feed the model, backends that serve it, infrastructure that keeps it up at 3 a.m.
I co-founded Applied AI, where I lead commercial AI deployments as CEO, and I've run Otherland Labs since 2017, delivering AI and engineering work for enterprise clients and government organizations. Before that: speech-to-text at Wasabi, media ML at GrayMeta, nationwide-scale ETL at DIRECTV.
What I'm building
-
FormFlow case study
Multi-brand form administration with AI image-authenticity verification for a major Polish food & beverage group — three independent checks per submitted photo. Designed, built, and deployed end to end at Applied AI.
Read the case study → -
CityChat in beta
A city guide you talk to — conversational discovery, camera-based landmark search, and a map built for wandering, grounded in curated open data. iOS first.
About the project → -
Applied AI operating
AI engineering company I run with my brother. From infrastructure to intelligence: cloud-native deployments, Go services, custom ML — the full stack, so AI works in production.
appliedai.dev ↗ -
Homelab AI cluster always on
A 4-node K3s cluster spanning home and office over a private mesh — GPU workstation serving self-hosted LLMs, embeddings, speech-to-text, and OCR. My personal AI stack runs on hardware I can touch.
k3s · self-hosted inference
Twelve years, shipped in order
-
AI Engineer & CEO · Applied AI
LLM customer-service agents, production RAG pipelines with vector search for enterprise knowledge bases, SaaS NLP products, LLM fine-tuning, and computer-vision quality control. Building the company alongside the systems.
-
Machine Learning Engineer · Wasabi Technologies, Boston
Optimized multilingual speech-to-text — +30% accuracy, +50% inference speed. Built diarization and LLM summarization pipelines; deployed Gemini, PaliGemma, and Phi-3 for automatic Q&A. Led CV projects: object recognition, OCR, brand detection.
-
Senior Software Engineer · GrayMeta, Los Angeles
Machine-learning algorithms and Go-based media processing at enterprise scale.
-
Software Engineer · DIRECTV
Designed and optimized large-scale ETL in Go, deployed across the nationwide U.S. market.
-
Software Engineer · Overlooked Inc.
Microservices Go backend end-to-end; application architecture strategy and development guidelines.
-
Founder · Otherland Labs, Przemyśl
AI/ML and engineering delivery for AT&T/DIRECTV, NBC Universal, and government organizations. Full-stack AI products and Go microservice backends for B2B clients across US and EU markets.
-
Freelance Software Developer
Web and backend work for clients worldwide — JavaScript, Angular, Go.
Tools of the trade
LLMs & AI
- Gemini · Azure OpenAI
- PaliGemma · Phi-3 · Whisper
- RAG · embeddings · vector search
- Fine-tuning · prompt engineering
- LangChain · AI agents
ML & data
- PyTorch · TensorFlow
- scikit-learn · pandas · NumPy
- Computer vision · NLP
- SQL · ETL · big-data pipelines
Engineering
- Python · Go
- Kubernetes · Docker
- CI/CD · cloud-native
- Microservices in production
Have a system that needs to think?
I'm open to interesting AI engineering problems — through Applied AI for company engagements, or directly.