How ruturaj-007/zyra-gnani works
Zyra is a browser-based, adaptive voice-interview product: a candidate sets up an interview, answers spoken questions, and reviews scores, reasoning, and latency in a results timeline. The backend handles voice turns, interview planning and evaluation, checkpointing, resume, and trace retrieval; the frontend provides setup, live-interview, and results experiences. The sampled backend excerpts support its calls and persistence paths. Frontend interactions and some integrations are documented in the README but their implementation details were not sampled.
Main entry points: api.ts API client · main.py FastAPI application · page.tsx Results page · voice.py Voice endpoints
Components 22
Candidate experience
- Interview setup
frontend/app/(root)/page.tsx - Interview page
frontend/app/(root)/interview/[id]/page.tsx - Voice interaction
frontend/components/Agent.tsx - Audio capture
frontend/lib/recorder.ts - API client
frontend/lib/api.ts - Candidate identity
frontend/lib/candidate.ts - Results page
frontend/app/(root)/interview/[id]/page.tsx - Reasoning timeline
frontend/components/TraceTimeline.tsx - Score chart
frontend/components/RadarChart.tsx - Latency panel
frontend/components/LatencyPanel.tsx
Interview API
- FastAPI application
backend/app/main.py - Voice endpoints
backend/app/routes/voice.py - Resume endpoint
backend/app/routes/resume_interview.py - Trace endpoint
backend/app/routes/trace.py - Speech services client
backend/app/gnani.py - API data models
backend/app/schemas.py
Interview intelligence
- Interview orchestration
backend/app/interview_core.py - Planning and evaluation
backend/app/agent.py
Persistence and results
- Neon Postgres
backend/app/db.py
Other components
- Candidate
- Groq LLM
- Gnani speech APIs
Connections 27
- Candidate → Interview setup: sets interview
README.md - Interview setup → API client: submits setup
- Candidate → Voice interaction: speaks answers
README.md - Voice interaction → Audio capture: records audio
- Voice interaction → API client: sends voice turns
- API client → FastAPI application: calls API
README.md - FastAPI application → Voice endpoints: registers routes
backend/app/main.py - FastAPI application → Resume endpoint: registers routes
backend/app/main.py - FastAPI application → Trace endpoint: registers routes
backend/app/main.py - Voice endpoints → Interview orchestration: handles interview
backend/app/routes/voice.py - Voice endpoints → Speech services client: transcribes and speaks
backend/app/routes/voice.py - Interview orchestration → Planning and evaluation: plans and evaluates
backend/app/interview_core.py - Interview orchestration → Neon Postgres: reads and writes
backend/app/interview_core.py - Planning and evaluation → Groq LLM: requests LLM
README.md - Speech services client → Gnani speech APIs: calls speech APIs
backend/app/gnani.py - Resume endpoint → Neon Postgres: reads checkpoint
backend/app/routes/resume_interview.py - Trace endpoint → Neon Postgres: reads trace data
backend/app/routes/trace.py - Trace endpoint → API client: returns results
README.md - API client → Results page: provides trace
- Results page → Reasoning timeline: shows reasoning
- Results page → Score chart: shows scores
- Results page → Latency panel: shows latency
- Candidate identity → API client: identifies candidate
- Voice endpoints → API data models: validates payloads
backend/app/routes/voice.py - Trace endpoint → API data models: returns trace model
backend/app/routes/trace.py - Resume endpoint → API data models: returns resume model
backend/app/routes/resume_interview.py - FastAPI application → Neon Postgres: initializes database
backend/app/main.py