QLoRA Fine-Tune for a Customer-Support Domain Assistant
Overview
What this challenge is about.
You receive 8,000 anonymized support ticket pairs (question -> agent response), the company's product documentation (around 600 pages), and a strong RAG baseline already running in production. Fine-tune a 13B base model with QLoRA on a single rented A100 24GB. Evaluate the QLoRA model and the RAG baseline on a held-out 300-ticket test set with rubric-based grading (correctness, on-tone, completeness, safety). Report wins/losses per category and the dollar cost per training run. Write a 2-page memo on the conditions under which fine-tuning beats RAG-only for this use case.
The Brief
What you'll do, and what you'll demonstrate.
Compare a QLoRA fine-tuned assistant against a strong RAG baseline on a customer-support task and identify when fine-tuning is worth the maintenance cost.
Earning criteria — what you'll demonstrate
- Run QLoRA fine-tuning on a consumer-class GPU
- Compare fine-tuning to a strong RAG baseline fairly
- Build a rubric-based LLM evaluation for product-shaped outputs
- Reason about the long-term maintenance cost of a fine-tuned model
Program Fit
Where this fits in your program.
Sharpens the same skills your degree expects you to demonstrate.
Skills
Skills you'll demonstrate.
Each one shows up on your verified credential.
Careers
Roles this prepares you for.
Real titles. Real skill bridges. Pick the one closest to your trajectory.
AI Engineer
Comparing fine-tuning vs. RAG on a real product task and writing the maintenance-cost memo is exactly the day-one work of an AI engineer at any B2B SaaS deploying LLMs.
This challenge sharpens
- qlora
- rag
- fine-tuning
Machine Learning Engineer
Running QLoRA training on constrained GPUs and reporting honest evaluation results is core MLE work for any LLM team.
This challenge sharpens
- qlora
- pytorch
- huggingface
NLP Engineer
Designing rubric-based evaluation for product-shaped LLM outputs is the NLP-engineer skillset for support and assistant products.
This challenge sharpens
- llm-evaluation
- fine-tuning
- rag