Consultant for data-driven AI solutions

I support companies not only to understand AI processes but to use them productively. My focus is on implementing concrete, business-oriented solutions with modern LLM and agent technologies. The following example projects show typical use cases where I achieved efficiency, automation and new value creation:

Automated email classification and workflow triggers Development of a system for intelligent detection and handling of incoming emails (e.g. applications, invoices, support cases). Tech: OpenAI GPT-4, LangChain, FastAPI, Supabase, webhooks

AI-powered contract assistant for a LegalTech startup Implementation of an LLM-based assistant for automatic analysis, categorization and extraction of relevant data from legal documents. Tech: Llama 3, Pydantic, LangGraph, Pinecone (RAG), Docker

Internal knowledge search with RAG architecture Building a retrieval-augmented generation solution for a mid-sized company to make internal documentation, FAQs and manuals accessible via an AI interface. Tech: vLLM, FAISS, LangChain, HuggingFace, OpenSearch

Edge-capable anomaly detection in industry Development of a locally deployable vision-language system for visual quality control on edge devices (e.g. Jetson, A100). Tech: Vision Transformer, LLaVA, Torch, Nvidia TensorRT, FastAPI

Agent system for automated report generation Introduction of an agent framework that regularly aggregates data (e.g. from CRM, JIRA, calendar), prioritizes it and turns it into weekly management briefings. Tech: LangGraph, PandasAI, LlamaIndex, REST, Zapier API

LLM-powered matching system for a recruiting platform Design of a system that matches applications with job profiles, calculates scores and automatically forwards suitable profiles to recruiters. Tech: OpenAI Embeddings, LangChain Agents, FastAPI, Postgres

Delivery of data-driven solutions focusing on predictive analytics, cloud data platforms and business enablement in the Azure environment:

Predictive maintenance system Development of a forecasting system for machine failures using Azure and Python. Building high-performance ETL workflows in Azure Synapse; result: 20% fewer unplanned outages and optimized maintenance planning through predictive models.

Enterprise Analytics Platform (EAP) Design and implementation of an enterprise-wide analytics platform on Azure, connecting various data sources. Automated pipelines and visualizations (Power BI, Tableau) enabled 30% faster operational decision-making.

Cloud data migration & reporting Leading the migration of legacy data to the Azure cloud, including redesigning Power BI reports. Result: 40% faster data availability and better responsiveness in day-to-day operations.

Sentiment analysis for customer feedback Development of an NLP-based analytics tool with Python and Azure Databricks to evaluate customer feedback. The insights were made available through Power BI; result: 25% increase in customer satisfaction through targeted actions.

Development of data-driven systems to meet regulatory requirements, market analysis and risk assessment in the financial sector:

Data warehouse customization for compliance Modernization of the bank's internal data warehouse using Excel, VBA, SQL (DB2) and Java to meet new regulatory requirements. Optimization of ETL processes and reporting. Result: full compliance, consistent data flows and higher reporting quality across departments.

AI-based sentiment analysis Development of a tool for real-time analysis of market sentiment from financial news using Python, TensorFlow and Dash. Result: faster reactions to market changes, improved risk assessment and automated action recommendations for investment controlling.

Stress test platform for risk simulations Building a comprehensive platform for running stress scenarios and data quality analysis with Python, SQL, Power BI, Java and React. Result: 30% higher responsiveness in crises, improved forecasting accuracy and solid decision support in risk management.