Nik S.

Based in Spain (EU, CET), GDPR requirements and EU data residency. I build AI and RAG systems that run in production at real scale: tens of millions documents, thousands of users. I deliver a working prototype on your own data, at a fixed price, built to grow into production. Most recent work 2025–2026: • Hospitality: built components of a revenue system, with the AI part running about 3× cheaper than OpenAI or Claude APIs; the client kept me on for long-term maintenance. • E-commerce: data pipeline, RAG-based AI assistants and search, used by around 1,500 users. • Hospitality: built from scratch a document-processing pipeline handling millions of documents, with optimized self-hosted GPU inference. What I do • AI assistants and search over your internal documents (RAG) • LLM/AI deployment: API models or self-hosted open-source models, fine-tuning and quantization • Document processing: OCR and extraction into structured data • Bare-metal and hybrid infrastructure: GPU servers, colocation, cloud bursting, to cut cloud bills • Cloud and DevOps on AWS, GCP and Azure, including cost optimization • Data pipelines and search infrastructure behind it all Background: 12 years building backend, data and cloud systems, and hands-on with hardware since 2013: I ran my own GPU fleet 24/7 for four years (2016–2020) and built servers for colocation. So I can tell you when self-hosting beats the cloud, and how to keep GPUs running reliably. Stack: Python, FastAPI, Django, Java/Spring · OpenAI, Hugging Face, TGI, Ollama · pgvector, Qdrant, Milvus, Elasticsearch, MongoDB, PostgreSQL · AWS (EC2, RDS, S3, Athena, Textract), GCP, Azure · Docker, Kubernetes, Terraform, Ansible · n8n Tell me what data you have and what you want it to answer, and I'll come back with an approach and a fixed-price estimate.