LLM Hosting & Knowledge Graph App

I have a cloud server already spun up on AWS/GCP and need an experienced machine-learning engineer to take it from empty shell to fully-functioning internal knowledge assistant. The job is threefold:

• Deploy and fine-tune a suitable large-language model on the existing instance, ensuring it can scale for concurrent queries from my team.
• Build a lightweight, browser-based conversation interface (Python/FastAPI or Node/Express on the back end, React/Vue on the front end are fine) that lets employees chat with the model, store each dialogue, and authenticate with our SSO.
• Transform every conversation into a continuously updated knowledge graph so we can run graph-based analytics later. Neo4j, Amazon Neptune, or any other graph database you prefer is acceptable as long as the data schema is documented.

Acceptance criteria
1. I can open the web app, start a new session, and get streaming answers in under two seconds.
2. Each turn of the dialogue appears as nodes and relationships in the graph store and is queryable with Cypher or an equivalent language.
3. Deployment scripts (Terraform, Ansible, or CloudFormation) let me recreate the stack on a fresh server.
4. README covers model specs, environment variables, and how to add new data sources.