Dennis Joseph

Development of Multi-Agentic AI Workflow – LangGraph, AutoGen, KnowledgeGraph, GraphRAG, MCP, Neo4j, Embeddings • Cloud Deployment and Versioning – Docker, Azure Container Apps, Azure DevOps CI/CD. • Azure Infrastructure – VNets, Application Gateway, Cosmos DB, Azure Bolb Storage, PostgreSQL, Redis, Container Registry, OpenID, and RBAC. • Debugging and Monitoring – Azure Monitor with KQL alerting, Grafana dashboards, and n8n for incident detection and GitLab issue creation.

LLM-guided RL Framework – Franka Panda arm, 4,096 parallel envs in NVIDIA Isaac Lab; PPO-based pick-and-place; 66% higher success rate over baseline RL. • Agentic LLM Pipeline – LangChain agents analysing training metrics for Hyperparameter tuning, Reward shaping, and termination adjustments. • System Architecture – Three-server setup (RL, LLM, dashboard) with human-in-the-loop parameter updates, real-time monitoring, and safe edge deployment

Containerization & Version Control – Docker, Docker Compose with multi-stage builds; Git patch management and GitHub Actions CI/CD for repeatable deployments. • Network Authentication – FreeRADIUS configured inside Docker containers for reliable and auditable access control. • Signal Validation & ECU Extraction – Unit tests using ARXML and CANoe log files; Python and ODXTools to automate PDX parsing and export structured ECU signal data.

ECU Flashing & Integration – DTS9 flashing environment setup, flash package management, and rest bus simulation using Vector CANoe for isolated ECU testing. • Validation & Debugging – Smoke tests post-integration and low-level microcontroller debugging with Trace32 for deviation analysis and root cause identification. • Tooling & Automation – Python and SQL tools for flash requirement generation, release artifact tracking, and process traceability via REST APIs.