AI Engineer / ML Engineer / AI Automation Engineer — LLM Fine-Tuning + RAG (Python)

**Description:** We're building a domain-specific AI assistant and need an AI Engineer / ML Engineer who's equally comfortable with model fine-tuning, RAG pipelines, and automation. This is a hands-on Python role — not prompt engineering. **What we need:** **Fine-tuning:** - Build a clean training dataset from our raw domain documents (we have thousands of files — PDFs, HTML, transcripts) - Data extraction, cleaning, and formatting into instruction/response pairs - Fine-tune an open model (Llama, Mistral, or similar) — LoRA/QLoRA approach is fine - Run evals to prove the fine-tuned model actually beats base + RAG on our domain **RAG layer:** - Build a retrieval pipeline over the same corpus — chunking, embeddings, vector store - Combine fine-tuned model + RAG so we get domain fluency *and* fresh/citable answers - Handle "I don't know" gracefully and cite sources **Automation & engineering:** - All in Python — clean, reproducible code - FastAPI endpoint to serve the assistant - Automate the data → train → eval → deploy loop so we can retrain easily - Lightweight eval harness so we can measure improvements over time You should have actually shipped a fine-tuning project before, not just followed a tutorial. Please share a GitHub repo, notebook, or writeup of past dataset-building / fine-tuning / automation work. **Skills:** AI Engineer, ML Engineer, AI Automation Engineer, Python, LLM Fine-Tuning, LoRA/QLoRA, RAG, Hugging Face, Vector Databases, FastAPI, PyTorch