Dr. Dimitar Denev

Semantic Systems Architect · AI Engineer · Head of Product

Munich, Germany
dimitar@denev.me
denev.me

Architect and AI engineer with 20 years building semantic search, retrieval, and intelligent information systems — from large-scale NLP pipelines and hybrid AI architectures to LLM deployment and AI-native workflows. PhD in Computer Science (Max-Planck-Institute for Informatics). Head of Product at moresophy GmbH, leading ContextSuite and DAPHY®, a patent-pending hybrid AI architecture for context-aware enterprise AI. Currently exploring agentic workflows, local LLM systems, MCP integrations, and human-in-the-loop interaction patterns.

Professional Experience

Software Architect, AI Engineer & Head of Product Oct 2012 – present
moresophy GmbH, Munich · AI platform company · BMWK-funded R&D, BSFZ seal 2021 & 2024
Semantic Search & Retrieval (2012–present)
  • Architect of a semantic search engine on Elasticsearch (migrated from Solr Cloud) — entity filtering, synonym handling, embedding-based similarity, multi-signal relevance ranking, metadata-aware retrieval
  • Built metadata enrichment pipelines, trainable AI filters, embeddings-driven classifiers, and context construction for downstream retrieval
  • Designed REST services for word-embedding classifiers/taggers and transformers for text summarisation
ContextSuite & Hybrid AI (2016–present)
  • Architected ContextSuite — retrieval and analytics platform with multi-source data integration, unified APIs, and information architecture for enterprise data
  • Designed DAPHY® (Data-driven Prompting with Hybrid AI) — context-aware orchestration layer combining analytical AI (NER, classifiers, embeddings) with generative AI; retrieval-driven prompting from live enterprise data; patent-pending hybrid intelligence architecture; up to 80% LLM resource reduction
  • Head of Product for ContextChat — enterprise conversational AI built on ContextSuite + DAPHY®
  • Architected ContextAnalytics — automated audit platform delivering 100% dataset coverage (zero sampling bias) for qualitative + quantitative analysis
  • Production NLP services: thematic classification, NER, key phrase extraction, persona analysis, risk classification
LLM Deployment & AI-Native Workflows (2023–present)
  • Exploring AI-native development workflows and agentic runtime architectures — local LLMs (Ollama, vLLM), MCP integrations, Claude Code, observability tooling, human-in-the-loop interaction patterns
  • Deployed data-sovereign, GDPR-compliant LLM infrastructure (on-premise + cloud on Kubernetes) — eliminates third-party data exposure; purpose-built for regulated industries such as banking and healthcare
  • Driving organisational adoption of AI-assisted software development across engineering and product teams
Research Assistant 2006 – 2012
Max Planck Institute for Informatics, Saarbrücken (2008–2012) · FIT Fraunhofer Institute, Sankt Augustin (2006–2008)
  • MPI: Designed crawling strategies for high-quality Web archiving; implemented prototype on Heritrix 3.0
  • Fraunhofer: Systems for content adaptation and Web standards compliance; W3C Web Accessibility Initiative contributor

Skills

Retrieval & Semantic Systems: Elasticsearch, OpenSearch, Solr, vector search, embeddings, RAG, relevance ranking, metadata enrichment, context engineering
AI & LLMs: DAPHY® hybrid AI, LLM deployment (Ollama, vLLM), MCP servers, Claude Code, prompt/context engineering, LangChain, LangGraph, OpenAI API, Claude API
NLP: NER, classifiers, embeddings, Hugging Face, spaCy, NLTK
Python: FastAPI, Flask, Pandas, NumPy, scikit-learn, Pydantic
Infra: Kubernetes, Docker, Helm, ArgoCD, Nginx, Traefik, GitLab CI, Cloudflare Workers, Azure
Databases: PostgreSQL, MySQL, MongoDB, Redis
Languages: Java, Groovy, Python, JavaScript/TypeScript, REST, GraphQL

Education

PhD in Computer Science
Max-Planck-Institute for Informatics · 2008–2012
MSc in Computer Science
RWTH Aachen, Germany · 2003–2006
BSc in Computer Science
Sofia University, Bulgaria · 1998–2002

Awards

  • ContextSuite — Best Media Solution 2022
  • VLDB Journal Special Issue — Best Papers VLDB 2009
  • John M. Slatin Award, W4A Conference 2008
  • DAAD Scholarship 2004/05

Publications

  • Denev et al.: SHARC Framework for Data Quality in Web Archiving, VLDB Journal, 2011
  • Velasco, Mohamad, Denev: Web Compliance Engineering Framework, W4A 2008