Self-hosted AI stacks, ready to deploy
Run AI on your own infrastructure: self-hosted RAG pipelines, chatbots and agents with Docker Compose, Helm and Terraform configs, plus practical guides.
Learn from use cases, deploy with stacks: browse AI stacks, explore the use case library or read the docs.
Popular AI stacks
- LLM Chatbot (chatbot) – Private chat UI (Open WebUI) for local models served by Ollama
- AI Agent Framework (agent) – Visual builder for LLM agents based on Flowise, with local models through Ollama and PostgreSQL storage
- Embedding Service (embedding) – Self-hosted text embedding API with Hugging Face Text Embeddings Inference (TEI), optional reranker
- RAG Pipeline (rag) – Ask questions about your own documents: Qdrant, Ollama and a small ingest and query API
Latest AI use cases
- Fraud Detection for E-Commerce Payments – Online shops and payment providers must decide within the payment flow whether a card-not-present transaction is legitimate. Fraud often shows up as patterns rather than single signals: a burst of… (1 solution)
- Medical Image Classification for Research and Education – Research and education use case. The results are not a medical device and must not be used for diagnosis or other clinical decisions. Researchers and students who work with X-ray, CT or MRI data… (1 solution)
- Crop Health Monitoring with Drone Imagery – Farmers and agronomists cannot walk every part of a large field often enough to spot problems early. Drones with multispectral cameras can capture the whole field in one flight, and vegetation… (1 solution)
- Privacy-Preserving Crowd and Traffic Counting – Cities, transport operators and venue managers need numbers about how public spaces are used: how many pedestrians and cyclists pass a street, how full a square or platform is, how much traffic of… (1 solution)
- Document Processing & Information Extraction Pipeline – Many teams still type data from PDFs, scans and photos into their systems by hand: invoices, contracts, forms. The goal is a pipeline that reads these documents automatically and returns structured… (2 solutions)
- Inventory Demand Forecasting with Dynamic Pricing – A retailer wants to predict demand per product and store and adjust prices automatically, so that popular items do not sell out and slow items do not pile up in stock. Forecasts have to account for… (3 solutions)
- Customer Support Email Classification & Auto-Response System – Support teams receive hundreds of emails a day across billing, technical and sales topics. Sorting them by hand delays the first response and sends emails to the wrong team. The goal is a system… (3 solutions)
- Smart Email Auto-Reply System – Small businesses and freelancers answer the same questions by email every day: prices, availability, order status, opening hours. An email assistant should sort incoming mail into categories and… (1 solution)
From the blog
- From Zero to Hero Part 1: A Short History of Machine Learning – How machine learning got from early neural network models and checkers programs to today's deep learning systems.
- From Zero to Hero Part 2: What is Machine Learning? – How machine learning differs from traditional programming, and why that shift in thinking matters.
- From Zero to Hero Part 3: Understanding Regression vs Classification – Predicting numbers versus predicting categories: the two core problems behind most machine learning applications.
- From Zero to Hero Part 4: The Four Machine Learning Paradigms – Supervised, unsupervised, semi-supervised and reinforcement learning: the four basic approaches behind most machine learning systems.