Hi, I'm
Data Scientist & AI Expert
I design and deploy intelligent systems — from Vision AI and Graph RAG pipelines to enterprise-grade AI platforms with governance, guardrails, and on-premise sovereignty.
I'm a Data Scientist and AI Expert with hands-on experience building production-grade AI systems across industries. My work spans the full AI lifecycle — from exploratory data analysis and model development to deploying scalable, secure AI infrastructure.
A particular focus of mine is enterprise AI strategy and governance: designing on-premise AI platforms using tools like Ollama, implementing authentication layers, content guardrails, and responsible AI policies that meet enterprise compliance requirements. I bridge the gap between cutting-edge AI research and real-world business deployment.
Whether it's a multimodal Vision AI pipeline, a Graph RAG knowledge system, or an end-to-end LLM platform — I build AI that is robust, explainable, and trustworthy.
Download CVObject detection, image classification, OCR, and video analytics pipelines using state-of-the-art models (YOLO, ViT, SAM).
Knowledge graph construction, entity extraction, and retrieval-augmented generation over structured graph databases for complex reasoning.
Custom chatbot development with fine-tuned LLMs, prompt engineering, multi-turn conversation management, and tool-use integrations.
Semantic search infrastructure using embeddings and vector databases (Chroma, Qdrant, pgvector, Weaviate) for scalable knowledge retrieval.
Enterprise AI roadmaps, responsible AI frameworks, policy design, guardrails, compliance controls, and AI risk assessment.
End-to-end private AI deployments using Ollama, open-weight models, SSO/auth integration, and air-gapped infrastructure for data-sensitive environments.
Interactive fruit classifier fine-tuned on 131 categories using EfficientNet-B0 and served at sub-second latency via ONNX Runtime — no PyTorch required at inference time.
Snap a selfie, pick a style — Ollama Vision validates your photo, rembg removes the background, then Flux Dev FP8 img2img transforms you into a comic illustration in seconds.
Agentic AI that uses the Model Context Protocol to search, read and create Confluence pages via a tool-use loop — answering natural language questions grounded in real knowledge base content.
AI agent that reads architecture docs from Confluence, then auto-generates a complete Jira backlog — Epic, User Stories with Acceptance Criteria, and Xray Test Cases — from a single plain-language requirement.
Agentic pipeline that generates platform-specific copy and hashtags, optionally renders a matching image via ComfyUI Flux Schnell, then routes through a Human-in-the-Loop approval step before publishing to Facebook / Instagram via the Meta Graph API.
Select from celebrity and character voices, then convert any text or your own recorded speech into that voice — entirely on local infrastructure using Kokoro TTS and Whisper STT.
Live two-way voice conversation with an animated portrait avatar powered by LiveKit Agents, Ollama LLM, Kokoro TTS, and Whisper STT — entirely local, no cloud APIs required. MediaPipe-detected lip sync animates a real photo in real time.
Voice bot that joins Microsoft Teams meetings via Azure Communication Services — transcribes live audio with Whisper, conducts interactive Q&A via Ollama + Kokoro TTS, and posts a structured summary or action item list to the meeting chat through Microsoft Graph.
Private, air-gapped AI platform built on Ollama with LDAP/SSO authentication, role-based access control, prompt guardrails, audit logging, and a multi-model API gateway for internal teams.
Production chatbot with dynamic RAG, intent routing across domains, human escalation logic, and a fine-tuned domain adapter. Deployed via Azure OpenAI with fallback to local Mistral.
Multi-tenant semantic search over 10M+ documents using hybrid dense/sparse retrieval (Qdrant + BM25), re-ranking, and a React-based search UI with real-time highlighting.
Designed and implemented an enterprise AI governance framework including model risk assessments, bias auditing pipelines, explainability reporting, and a compliance dashboard aligned with EU AI Act requirements.
Interested in AI strategy consulting, a speaking engagement, or building something together? Let's talk.
Thanks for reaching out. You'll receive a confirmation email shortly — I'll get back to you as soon as possible.
Data privacy (DSGVO/GDPR): By submitting this form your name, email address, and message are transmitted to the website operator solely for the purpose of responding to your enquiry. No data is passed to third parties. Data is stored only as long as required to process your request and is then deleted. You may request deletion at any time by contacting timo@racke.org. Legal basis: Art. 6 (1) lit. b GDPR (pre-contractual measures) or Art. 6 (1) lit. f GDPR (legitimate interest).
Responses generated by a local Ollama instance via RAG. May be inaccurate.