feat: add NVIDIA NIM as a BYOK LLM provider (JEF-63)

[mankatcheung]

Summary

Adds NVIDIA's NIM inference API (build.nvidia.com) as another selectable provider in the existing multi-provider BYOK system, alongside OpenAI, Anthropic, Google AI, OpenRouter, Mistral, Groq, xAI, DeepSeek, and Custom.

NIM's hosted endpoint (https://integrate.api.nvidia.com/v1/chat/completions) implements OpenAI's /chat/completions wire format, so this reuses the existing OpenAICompatibleLLMProvider via a new providerRegistry.ts entry — no new provider class needed, following the same pattern already used for Mistral/Groq/xAI/DeepSeek.

• constants.ts: LLM_PROVIDER.NVIDIA + LLM.NVIDIA_API_URL/NVIDIA_DEFAULT_MODEL
• providerRegistry.ts: new registry entry
• account.tsx: added 'nvidia' to the provider Zod enum and the dropdown options list

Default model is meta/llama-3.1-8b-instruct (small, fast — consistent with the other providers' lightweight defaults); users can override it like any other provider via the existing model field.

Ref: Linear JEF-63.

Test plan

• pnpm --filter @job-finder/api typecheck / pnpm --filter @job-finder/web typecheck — clean
• pnpm --filter @job-finder/api lint / pnpm --filter @job-finder/web lint — clean
• pnpm --filter @job-finder/api build / pnpm --filter @job-finder/web build — clean
• pnpm --filter @job-finder/api test — 855/855 passing. providerRegistry.test.ts generically iterates every LLM_PROVIDER value and asserts each has a registry entry, so it covers NVIDIA automatically without changes.
• pnpm --filter @job-finder/web test — 148/148 passing.
• Manually verified end-to-end against a live dev server + real SQLite DB: registered a user, called saveLlmApiKey with provider: "nvidia" directly over GraphQL and confirmed llmKeyStatus read it back correctly, then repeated the flow through the actual browser UI via Playwright — selected "NVIDIA NIM" from the Account page dropdown, saved a key, and confirmed via screenshot that the page shows "AI features are enabled using NVIDIA NIM."

🤖 Generated with Claude Code

https://claude.ai/code/session_01N2PBmsuzPhrmNnfZf6C3BM

Summary by CodeRabbit

• New Features
  • Added NVIDIA NIM as a supported AI provider.
  • Added NVIDIA NIM API configuration and a default model.
  • Added NVIDIA NIM to provider selection and API-key validation in account settings.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro Plus

Run ID: 89821f0a-0f0e-4072-af47-1250d015e69a

📥 Commits

Reviewing files that changed from the base of the PR and between 8e9035b and 73b8527.

📒 Files selected for processing (3)

• apps/api/src/constants.ts
• apps/api/src/infrastructure/llm/providerRegistry.ts
• apps/web/src/routes/_authenticated/account.tsx

---

Walkthrough

The API adds NVIDIA NIM as an OpenAI-compatible LLM provider. The account settings accept nvidia and display “NVIDIA NIM” as a provider option.

Changes

NVIDIA NIM provider

Layer / File(s)	Summary
Provider constants and registry apps/api/src/constants.ts, apps/api/src/infrastructure/llm/providerRegistry.ts	The API defines the NVIDIA NIM endpoint and default meta/llama-3.1-8b-instruct model. The registry creates an OpenAI-compatible provider with these values.
Account provider selection apps/web/src/routes/_authenticated/account.tsx	The validation schema accepts nvidia. The provider list displays “NVIDIA NIM”.

Estimated code review effort: 2 (Simple) | ~10 minutes

Sequence Diagram(s)

sequenceDiagram
  participant AccountSettings
  participant ProviderRegistry
  participant OpenAICompatibleLLMProvider
  participant NvidiaNIM
  AccountSettings->>ProviderRegistry: select nvidia with API key
  ProviderRegistry->>OpenAICompatibleLLMProvider: create provider with NVIDIA endpoint and model
  OpenAICompatibleLLMProvider->>NvidiaNIM: send chat-completions request

Loading

Possibly related issues

• mankatcheung/job-finder#175 — Adds NVIDIA NIM as an OpenAI-compatible BYOK provider across the API and account settings.

Possibly related PRs

• mankatcheung/job-finder#45 — Extends the provider constants, registry, and account provider selection.
• mankatcheung/job-finder#152 — Uses the provider constants and registry in the per-user provider factory.
• mankatcheung/job-finder#155 — Modifies the same provider registration and account selection paths.

Poem

A bunny adds a provider with care,
NVIDIA NIM now joins the lair.
Constants guide the model’s call,
The registry connects them all.
“NVIDIA NIM” appears in view,
With chat completions hopping through.

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly and concisely identifies the addition of NVIDIA NIM as a BYOK LLM provider, which is the main change.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch worktree-jef-nvidia-llm-provider

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[github-actions]

Preview deployments for this PR:

• Web: https://job-finder-126qkepec-mankatcheungs-projects.vercel.app
• API: https://job-finder-mg6latr52-mankatcheungs-projects.vercel.app