Environment
Client mentioned that we need a plug and play, easy to set up and use solution incorporating: Github Copilot Fabric Foundry
What “plug and play” should mean (proposal)
- One click to a working environment: a GitHub Codespace / dev container with VS Code, GitHub Copilot, HVE Core, the Azure and Fabric CLIs, and the starter repo pre-installed. No laptop installs beyond a browser.
- Pre-provisioned cloud: Foundry project and model deployments and a Fabric workspace already exist and are wired into the container’s settings
- Setup check before day 1 on the course platform: sign in, open Codespace, run one Copilot chat, hit one model endpoint
- One command per cohort to provision and tear down (infrastructure as code)
- Details of HVE install options: HVE Installation and Cohort Baseline
Requirements
| Need | For | Notes |
|---|---|---|
| GitHub org + Copilot seats | Github Copilot, HVE | Coding agent & org policies need Business/Enterprise |
| Codespaces / dev containers | All | Removes laptop-setup risk; pre-install HVE Core |
| Azure subscription + model quota | Foundry | Region with needed models; per-learner budget caps |
| Fabric capacity | Fabric | Pause when idle — biggest cost risk |
| Sample data + starter repos | Sample project (Questions) | Synthetic data only unless decided otherwise |
Principles
- Per-learner isolation where cheap, shared where expensive (e.g. one Fabric capacity, per-learner workspaces)
- Cost guardrails: budgets, alerts, auto-pause. Lab cost × 10,000 learners is a real line item → Scale and Capacity Model
- Version-pinned starter repos tagged per cohort
- Firm constraints (network, data, model usage) may limit what runs → ^q-026
Open
- Tenancy & cost: ^q-009. Ask Microsoft first; they may provide lab tenants/credits for a program they brand.
Setup Planning
See Environment and Setup Plan for the proposed pilot environment, learner preflight, version baseline, recovery paths, and unresolved provisioning decisions.