RAGAS Experts in Germany

• Acted as technical consultant to optimize multi-layer ensemble models combining ResNet, CNN-BiGRU-Attention, and XGBoost.
• Guided implementation of a Logistic Regression meta-learner to solve class imbalance problems, achieving 92.86% accuracy and 0.9644 AUC on PTB-XL and Chapman-Shaoxing datasets.

End-to-end responsibility for all products - concept, architecture, development, and production operation as the sole developer; in addition, customer meetings, proposals, and marketing.

Underwriting Copilot - AI assistant for industrial insurance (in production at customer sites)

• Supports underwriters in analyzing industrial insurance submissions - in production use at an industrial insurer.
• Framework-independent RAG architecture with Hybrid Search (BM25 + pgvector) across large, mixed document sets.
• 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 at customer sites)

• Automatically captures and analyzes menu 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.
• Full CI/CD pipelines (GitHub Actions), production cloud deployment, interactive dashboards (Dash).

Kaiki GEO Atlas - GEO platform (in production at customer sites)

• 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 extra LLM cost).

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 progress)

BankingGPT - AI assistant for complaint management in cooperative banking

• Security architecture at the core: no AI draft reaches the customer without human approval - the approval decision is in auditable code, not in the language model (monotonic: the model may escalate, never downgrade).
• Real agentic building blocks, each with its own boundary: the model chooses tools itself through an MCP server (read-only, allowlist, capped, fail-safe); sensitive cases are handed off via an open A2A protocol (JSON-RPC, Agent Card, message/send/tasks/get; client implemented by me) to a separate specialist agent (securities/law), which never lowers the review requirement (pinned by test).
• Evaluation-driven over ten analysis rounds; uncovered a security flaw through independent review and blind tests that nine automated runs had missed.
• Voice AI frontend, responding live: covered cases are answered in the conversation, sensitive ones escalate before generation; response latency < 7 s measured (local GPU STT/TTS).

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

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

• Genuinely agentic on LangGraph: ReAct agent with four tools and conversation memory - the model decides on its own whether to use the manual (RAG, Chroma), a knowledge graph (GraphRAG, Neo4j/Cypher - decodes warning lights), or a workshop/booking service.
• Human-in-the-Loop before the irreversible action: before every appointment booking, the graph pauses (interrupt) and gets the driver's explicit confirmation - the same approval-before-action discipline as in BankingGPT, in a different framework.
• Eval as CI gate: a three-part scorecard (RAGAS grounding + deterministic tool-routing accuracy + DeepEval safety: does the answer mention the warning first when there is 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.

• Led end-to-end product discovery and strategy engagements for early-stage founders (pre-seed) clients, defining product vision, OKRs, and go-to-market strategies and roadmaps for their cloud-based and AI-enabled solutions.
• Designed scalable product operating models (Agile/Scrum, backlog governance, KPI frameworks) as part of the partnership, to improve delivery predictability and reduce feature cycle time by up to 30%.
• Built scalable product roadmaps aligned to fundraising milestones, helping founders articulate product vision, product<>market fit, and traction clearly to investors.

From Crude to Crafted: Refining Participatory Design Data into Stakeholder-Ready Outcomes

• Architected a production Document AI platform using Retrieval Augmented Generation (RAG) over 1,500+ participatory design artefacts to answer historical project queries with grounded responses.
• Designed LLM evaluation combining RAGAS, custom evaluation metrics and human-in-the-loop (HITL) validation workflows to evaluate factual grounding, response quality, and prompt performance.
• Built a React, TypeScript, and D3.js frontend for interactive exploration of AI-generated insights.
• Implemented input layer LLM safety controls and Guardrails, including PII redaction and foul language filtering.

• Developed a multi-agent system connecting LangChain ZeroShotAgent with custom tools for live APIs and task automation.
• Built a FastAPI backend for Jira ticket creation, triage and assignment, auto classification of severity, deduplication, SLA setup, on-call rotation, bidirectional sync of status and comments.
• Added Slack alerts and RAG knowledge lookup with FAISS or pgvector to suggest fixes, optional PagerDuty escalation on policy breaches.
• Orchestrated agents with a router and a Celery plus Redis queue, retries with backoff, rate limits, idempotency keys, human in the loop approvals.
• Implemented guardrails and observability, prompt versioning, token and cost budgets, PII redaction, tool-use allowlists, timeouts, OpenTelemetry tracing, dashboards for accuracy and latency, deployed on Kubernetes with feature flags and canary rollouts.

• Build a sophisticated AI copilot to augment the students’ learning experience and provide AI-derived insights to professors.
• Build a multi-chain LLM system adapting to user needs at its own accord with a Weaviate vector DB based RAG system and evaluated it with Ragas.
• Build responsive react frontend, and backend systems handling auth, data management and auxiliary services as a RESTful API.
• Deployed and managed the app to the cloud in a production environment including the CICD via multi-stage deployment.

• Built an AI Research Assistant with RAG, LangChain, LangGraph, and OpenAI LLMs integrated with vector search; cut research time by 30%.
• Designed custom retrieval workflows with LlamaIndex, building a ReAct-style agent for dynamic chunking; improved query accuracy by 18%.
• Researched and optimized embedding strategies, reducing retrieval cost/query by 15%.
• Developed RAG evaluation frameworks using RAGAS and Langsmith with custom datasets; improved coverage by 40%.
• Fine-tuned LLMs (LLaMA 2 on Vertex AI with custom inference containers, dynamic batching, and quantization); reduced inference latency by 25%.
• Integrated AI agents in LangGraph with short-term & long-term memory (Mem0); increased task completion rate by 20%.
• Created schema-aware synthetic data generators; fine-tuned downstream models achieving +12% F1 score.

• Developed a RAG system to automate student queries with 96% accuracy, built using FastAPI and LangChain and deployed on the university server with Docker.
• Evaluated performance using RAGAS, comparing LLMs (Llama3.3, Llama3.1, GPT-4o-mini), vector embeddings, and various retrieval techniques within the RAG pipeline.
• Technical Skills: Python, FastAPI, Docker, AWS, LangChain, LangSmith, NLP, HTML, CSS