AutoEmbed

• Python
• LangGraph
• RAG
• LLM agents
• Ollama (local)
• Cohere (cloud)
• AUTOSAR ARXML

What it is

AutoEmbed is where my embedded background meets agentic AI. AUTOSAR projects are configured through large XML files (ARXML), and merging them with ordinary version control is painful — git and SVN compare text lines, not structure, so they raise false conflicts and miss real ones. AutoEmbed parses ARXML structurally and resolves conflicts by meaning: it classifies each change, validates it, and reports what actually happened.

What it does

• Semantic ARXML merge. Understands the configuration's structure and merges by intent — going beyond what line-based git/SVN diffs can do.
• Local RAG over a repo. Retrieval-augmented search across a codebase's code and docs for private, NDA-safe semantic search, code understanding, and debugging help.
• Runs local or cloud. Works with a local model (Ollama) for privacy or a cloud model (Cohere), driven by a LangGraph agent workflow behind a command-line interface.

Why it matters

It's a real problem I lived on Tier-1 automotive programs — resolving configuration and data-flow conflicts across component boundaries. AutoEmbed turns that experience into tooling, and it's proof that my embedded domain knowledge is exactly what makes an AI agent here useful: the value isn't the LLM, it's knowing what a correct merge means in AUTOSAR.