GraphRAG Experts in Germany

• This project involved extraction of intelligence data to support report writing for a company that provides geopolitical, global, commercial intelligence. The data have been gathered from a number of resources (interview transcripts, online data, internal documents), and then a knowledge base has been build from it. This was the basis of a complex RAG system, that was evaluated against a golden dataset. Agents have been used to find out the contradicting intelligence, the statements supporting each other, and to store back the generated knowledge.

Used: Python, RAG, LangGraph, LangChain, deepeval, MCP

Independent engineering practice focused on custom AI systems, production delivery, and fractional technical leadership.

Selected client engagement: Scalutions

• Role: Serve as fractional CTO and hands-on technical lead, responsible for the architecture and agentic infrastructure behind its managed B2B outbound operation.

• Product: Designed and built OutboundLoop, an agentic SDR operating system for research, qualification, personalized outreach, campaign management, human approvals, measurement, and continuous improvement.

• Scope: Own the full system lifecycle—from business processes and agent behavior to context design, model routing, integrations, evaluation, telemetry, reliability, cost control, and production operations.

End-to-end responsibility for all products – conception, architecture, development, and production operations as the sole developer; also customer discussions, proposals, and marketing.

Underwriting Copilot – AI assistant for industrial insurance (in production with a customer)

• Supports underwriters in analyzing industrial insurance submissions – in production at an industrial insurer.
• Framework-independent RAG architecture with Hybrid Search (BM25 + pgvector) across large, heterogeneous document collections.
• Two-stage evaluation and observability pipeline (code assertions + LLM-as-Judge) that makes answer quality, retrieval accuracy, and citation integrity measurable in a regression-safe way.

Kaiki Menu Analyzer – Data intelligence platform (in production with a customer)

• Automatically collects and analyzes food data from around 25,000 German restaurants.
• Scalable 7-container architecture (FastAPI, partitioned PostgreSQL, Redis/RQ) with LLM-supported extraction of structured data from PDF, HTML, and images.
• Complete CI/CD pipelines (GitHub Actions), production cloud deployment, interactive dashboards (Dash).

Kaiki GEO Atlas – GEO platform (in production with a customer)

• Measures brand visibility across five AI engines (ChatGPT, Gemini, Perplexity, Grok, Claude), each augmented with web search, orchestrated as a DAG workflow pipeline (Dispatcher → Sub-Workflows → Scoring → Report) with fail isolation.
• 6-container deployment (FastAPI, Celery, Redis, PostgreSQL); LLM cost estimation, PDF audit report, rule-based cross-signal insights (no additional LLM effort).

Data Pipeline & Analytics Platform – Competitive analysis in the automotive aftermarket

• Automated data pipeline with gap analysis algorithms and role-based access control; 230+ tests.
• Backend with FastAPI, PostgreSQL, SQLAlchemy.

Product development (actively in development)

BankingGPT – AI assistant for complaint management in cooperative banking

• Security architecture at its core: no AI draft reaches the customer without human approval – the approval decision is stored in auditable code, not in the language model (monotonic: the model may increase the level, but never lower it).
• Genuine agentic components, each with its own boundaries: the model selects tools itself through an MCP server (read-only, allowlisted, capped, fail-safe); sensitive cases are sent via the open A2A protocol (JSON-RPC, Agent Card, message/send/tasks/get; client implemented independently) to a standalone specialist agent (securities/law), which never lowers the review requirement (pinned by tests).
• Evaluation-driven across ten analysis rounds; an independent review and blind tests uncovered a security defect that nine automated runs had missed.
• Voice AI frontend, responding live: covered cases are answered during the conversation, while sensitive cases are escalated before generation; measured response latency < 7 s (local GPU STT/TTS).

Stack & production readiness: Python, pydantic-ai, FastAPI/Celery, PostgreSQL/pgvector, FastMCP, fasta2a, Docker; multi-tenant (physical vector isolation per tenant), encrypted PII, OWASP-LLM-reviewed, 275 tests, CI/CD; provider-portable (Ollama / EU-Cloud-Vertex).

After-Sales Assistant – agentic RAG/GraphRAG assistant based on public OEM manuals (automotive after-sales)

• Genuinely agentic with LangGraph: ReAct agent with four tools and conversation memory – the model decides itself whether to use the manual (RAG, Chroma), a knowledge graph (GraphRAG, Neo4j/Cypher – decoding warning lights), or a workshop/booking service.
• Human-in-the-loop before the irreversible action: before every appointment booking, the graph pauses (interrupt) and obtains the driver's explicit confirmation – the same approval-before-action discipline as in BankingGPT, in a different framework.
• Eval as a CI gate: three-part scorecard (RAGAS grounding + deterministic tool-routing accuracy + DeepEval safety: does the answer mention the warning first in the case of a critical warning?) blocks the pipeline; provider-agnostic (OpenAI/Azure/Anthropic), FastAPI with token streaming.

Stack: Python, LangChain/LangGraph, Chroma, Neo4j, RAGAS/DeepEval, FastAPI, Docker.

• Consultant for system architecture, AI agents & integration, coach for data & process logic, Graph-RAG approaches, security and data protection.
• On-premise AI solutions with high compliance and performance requirements.
• Architecture decisions, operational setup, strategic prioritization & deployment.
• Technologies: LiveKit JS SDK, LiveKit Agents, Web Audio API, JS, AudioWorklet, Loki, vLLM, Zscaler, Docker, Neo4j, MySQL, Python.
• Models: GPT-OSS 20B, Whisper large v3 turbo, Qwen3-TTS.

• Develop and demonstrate solutions using GenAI software like langchain, vercel ai sdk, copilotkit
• Work with customers to understand their challenges and provide the best solutions based on open-source data products
• Build RAG and GraphRAG solutions using Neo4j, lancedb, and Postgres
• Deploy a LLMOps platform using kubernetes, terraform, helmfile, Arize phoenix, mlflow
• Architect and build data pipelines using dbt, Trino, Spark, Iceberg, Airflow, ArgoCD, terraform, kubernetes
• Delivered user-centred technical strategy for Agriculture 4.0 and precision livestock farming, helping my client secure funding from Bpifrance
• Delivered a prospecting tool for a leading French solar carport installer, using geospatial computing (GIS), speeding up the sales process
• Built digital twin architecture for solar carports and EV chargers, making real-time monitoring and smart charging possible

• Lead engineer for the MatchMaker Toolbox (KNIME): Index Builder, Approximate Matcher, Character Mapper, license nodes
• Designed M|ARS (MatchMaker Agentic Retrieval System) — hybrid retrieval combining deterministic search + LLM tooling
• Developed internal RAG and search prototypes (MatchMaker + vector search + LLM)

• Further training in Snowflake and Google Looker
• Working with LLMs: local models (Llama, Mistral, Gemma, Phi, Qwen, DeepSeek, Bitnet, Flux, Whisper), OpenAI API, frontends (ollama, openwebui, loacalai)
• Inference methods: llama.cpp, vLLM, transformer
• Quantization, benchmarking, prompting
• LLM Agents (Tool/Function Calling, LangChain, LangGraph, MCP)
• Topics: attention, reasoning, chain of thoughts, RAG, GraphRAG, mlflow
• Cloud hosted: ChatGPT, Claude, Gemini

• Architected and deployed a production-scale GraphRAG system using Neo4j, embeddings, and multi-hop reasoning over 120M+ nodes, improving answer precision by 32%, reducing hallucinations by 41%, and lowering retrieval latency by 38%.
• Designed and implemented an enterprise agent ecosystem using Model Context Protocol (MCP), exposing internal APIs, databases, and services as secure callable tools for autonomous workflows and system integration.
• Designed and deployed a production LLM-based email routing agent using Microsoft Graph API, MCP, and Azure OpenAI, achieving 96% routing accuracy, reducing manual triage workload by 65%, and decreasing response times from 18 hours to under 4 hours.
• Implemented autonomous agent self-correction pipelines using iterative feedback loops (Ralph Wiggum), enabling reliable error detection, automated remediation, and production-safe execution.
• Developed a multimodal semantic search platform using multimodal LLMs and vector embeddings, enabling semantic discovery across 250k+ image and video assets and improving search recall by 48%.

• Designed GraphRAG system combining semantic vectors (Chroma) + knowledge graphs (NetworkX/Neo4j) for multi-hop Q&A on climate policy docs.
• Deployed Llama 3/Gemma via Ollama for $0-cost local inference; achieved ~95% entity-relation extraction accuracy.
• Built ingestion pipeline for PDFs + 48k-row CSVs; applied grouped median imputation and fixed data sparsity.

• Conducted stakeholder interviews and market analysis to identify challenges in secure genetic data use, translating insights into a product concept and AI graph-RAG prototype.