AI Interview System

What You Can Build

The AI Interview System powers real workflows built from the same primitives:

• ATS Integration — Push interview requests from your Applicant Tracking System via REST, using qualification statements instead of a skills taxonomy
• Webhook-driven pipelines — React to state changes (plan generated, approved, completed) via HMAC-signed webhooks to your registered endpoint
• PII-minimal workflows — Use candidateRef instead of candidateName/Email to avoid sending personal data; purge PII on demand after assessment
• Custom HITL interfaces — Build recruiter review tools against the approval workflow API, or enable auto-approve for fully automated flows
• Assessment automation — Receive structured AI assessments with qualification evaluations and pipe them into your HR systems

Start Building

The system is designed to scale with you from a single API call to production multi-tenant deployments.

• Quickstart — Create your first interview with a qualification-first request in under 5 minutes
• Integration API — Integrate your ATS or HR platform via REST using your platform API key
• Webhooks — Register your callbackUrl once per tenant to receive real-time state-change events
• HITL Workflow — Enable auto-approve for zero-touch flows or build custom recruiter interfaces

# Create your first interview — qualification-first, no PII required
curl -X POST https://mayaapi.teamcast.ai/api/v1/integration/interviews \
  -H "X-API-Key: your_api_key" \
  -H "X-Tenant-ID: your_tenant_id" \
  -H "Content-Type: application/json" \
  -d '{
    "candidateRef": "li_app_a1b2c3d4",
    "position": "Senior Engineer",
    "level": "SENIOR",
    "qualifications": [
      "5+ years production TypeScript experience",
      "Distributed systems design at scale"
    ]
  }'

# Response: { "runId": "run_...", "state": "VALIDATING_SKILLS" }

AI-Assisted Integration

Use our MCP (Model Context Protocol) server with Claude Code, Cursor, or Windsurf to get instant API integration help. Your AI coding assistant can search endpoints, read schemas, and access integration guides directly — no context-switching to docs.

Server URL

The server uses the Streamable HTTP transport. Add it to your MCP configuration:

{
  "mcpServers": {
    "teamcast-docs": {
      "url": "https://mayaapi.teamcast.ai/mcp"
    }
  }
}

Once configured, ask your AI assistant questions like:

• "How do I create an interview via the API?"
• "What fields does CreateInterviewDto require?"
• "Show me the webhook payload for interview.completed"
• "What permissions does the workflow approve endpoint need?"

Built for Production

AI Interview System runs in your infrastructure, not ours.

• Multi-tenant with row-level security at the PostgreSQL level
• Platform → Tenant hierarchy — one API key per platform, X-Tenant-ID per request
• Horizontally scalable — memory-based HPA on the Interviewer, CPU-based on Planner and Assessor
• Event-driven via Google Managed Kafka (SASL_SSL) — 100 partitions per audio topic
• Full audit trail via structured logging and Kafka event history
• HMAC-signed webhooks with exponential backoff retry
• Candidate PII purge API with immutable audit log
• Configurable retention windows per tenant

Production Performance

Benchmarked on the production GKE cluster — 2 interviewer pods, 3x n2-standard-8 nodes (24 vCPU / 96GB total). All numbers are from live cluster runs against real Kafka, Redis, Google STT, Vertex AI LLM, and Google TTS — no mocks.