LLM Fine-tuning Specialist — Customer Support Bot

We run a SaaS platform for e-commerce merchants. We want to build a fine-tuned support bot that handles 80% of incoming tickets automatically — refund requests, shipping questions, order status, and product FAQs. We have 18 months of historical support tickets (cleaned and labelled). We need someone to: 1. Evaluate whether fine-tuning vs. RAG vs. prompt engineering is the right approach 2. Run training experiments (likely Mistral or Llama fine-tuned via LoRA) 3. Build evaluation benchmarks 4. Deliver a deployed endpoint with documentation We are not looking for a chatbot builder who uses a no-code tool. This should be a proper ML project.