How sapthamikotian/athena works
Athena is a personalized technical-interview web app. A candidate is selected in the frontend and sent to the interview endpoint to start a session; subsequent answers go through the same endpoint, where retrieval and interview control shape the next question until coverage guardrails allow final feedback. The graph separates the web/API, interview orchestration, topic/session logic, and grounding/LLM services. The README documents the NVIDIA NIM integration and the 8-question/4-day floor; sampled code shows the routing and core calls. The frontend excerpt is partial, so UI details beyond its candidate loading and interview requests are omitted.
Main entry points: index.ts Interview API · handlers.ts Interview handlers · app.js Interview UI · vectorStore.ts Curriculum vector store
Components 14
Web and API
- Interview UI
frontend/app.js - Candidate profiles
candidates.json - Interview API
src/index.ts - Request validation
src/utils.ts
Interview orchestration
- Interview handlers
src/handlers.ts
Topic and session
- Session state
src/session.ts - Topic planning
src/topics.ts - Curriculum lookup
src/data.ts - Curriculum content
curriculum.json
Grounding and generation
- Grounding and memory
src/rag.ts - Curriculum vector store
src/vectorStore.ts - Interview and feedback generation
src/llm.ts
Other components
- Interviewer user
- NVIDIA NIM
Connections 18
- Interviewer user → Interview UI: uses
frontend/app.js - Interview UI → Candidate profiles: loads profiles
frontend/app.js - Interview UI → Interview API: starts and sends turns
frontend/app.js - Interview API → Interview UI: returns replies
src/index.ts - Interview API → Request validation: validates requests
src/index.ts - Interview API → Interview handlers: routes requests
src/index.ts - Interview API → Curriculum vector store: initializes store
src/index.ts - Interview API → Interview and feedback generation: initializes client
src/index.ts - Interview handlers → Session state: creates and advances
src/handlers.ts - Interview handlers → Grounding and memory: retrieves and updates memory
src/handlers.ts - Interview handlers → Interview and feedback generation: requests turns and feedback
src/handlers.ts - Session state → Topic planning: builds topic queue
src/session.ts - Topic planning → Curriculum lookup: looks up day details
src/topics.ts - Curriculum lookup → Curriculum content: loads curriculum
src/data.ts - Curriculum vector store → Curriculum lookup: embeds curriculum days
src/vectorStore.ts - Grounding and memory → Curriculum vector store: embeds and queries
src/rag.ts - Curriculum vector store → NVIDIA NIM: requests embeddings
src/vectorStore.ts - Interview and feedback generation → NVIDIA NIM: requests interview generation