Log Message Interpretation

Technology: Azure OpenAI, embeddings, vLLM, Guidance, GroupChat orchestration, Streamlit, Qdrant, Python

Project description: Built a pipeline that converts raw industrial log templates into Canonical Event Model (CEM) schemas for automated structured extraction. Implemented a two-phase architecture with entity extraction and canonical registry creation, followed by multi-agent schema generation and validation with fact extraction/verification. Added reviewed-blueprint retrieval with hybrid similarity scoring to improve schema quality, and delivered a Streamlit interface for manual review, schema edits, and quality metrics.

Contractual Intelligence

Industry: Automotive (automobile manufacturer)

Technology: Azure OpenAI, Azure AI Search, Azure Functions, Azure ML, Prompt Flow, Python, Document Intelligence

Project description: Contributed to a contractual intelligence solution that improved the efficiency of analyzing and comparing contract data with AI. Conducted data analysis to identify patterns across different contract types. Implemented and defined custom search indexes, indexers, skillsets, and data sources as the foundation for a hybrid search concept. Developed a Retrieval Augmented Generation (RAG) architecture with Azure AI Search and Azure OpenAI. Automated pre-processing tasks for each document, including content extraction and metadata enrichment using Document Intelligence. Built automated document cracking and chunking processes for storage in the related search index and implemented multiple Prompt Flow variations to handle different user prompt patterns.

Unified Customer Profile

Industry: Automotive (automobile manufacturer)

Technology: Azure Functions, Azure SQL, AKS, VM Scale Sets, Azure Firewall, Compute Gallery, Container Registry, Bicep, Azure Verified Modules, Azure Landing Zones, ASP.NET

Project description: Contributed to the development of a company-wide standard for core customer data. Built C# applications for multiple scenarios, including Azure Functions, ASP.NET APIs, Entity Framework integrations, unit testing, end-to-end testing, and performance analysis. Helped create a multi-stage environment to host all related services and enable seamless deployment without disrupting production workloads. Played a key role in developing the landing zone Infrastructure as Code repository and integrating it into multi-stage environment deployments.

Fruit and Vegetable Recognition

Industry: Retail

Technology: Azure Landing Zones, AKS, Helm, Private Networking, Azure Container Registry, Azure Verified Modules, Azure Edge Devices

Project description: Led the global rollout workstream as the development lead for an innovation project focused on automatic fruit and vegetable recognition at self-checkout terminals. Built a rollout-ready solution for edge devices and established the global CI/CD pipeline as the core foundation for product rollout. Developed Helm charts with sub-containers, hosted and versioned in Azure Container Registry, enabling independent version rollout per shop and preventing unplanned upgrades.