Hiring a voice AI engineer?
I'm Hassan Kazi. I lead the AI and real-time stack behind InCruiter's voice interviewers (1,000+ concurrent interviews, 42+ languages) and built its AI phone screening from scratch. Bengaluru, India · open to remote (IST).
What I can build for you
Real-time voice agents
LiveKit Agents over WebRTC: STT → LLM with tool calling → TTS, turn-taking and barge-in that hold up on real calls. See proof →
Phone-call agents
AI phone calls over Twilio and Exotel, orchestrated with Vapi; voicemail detection, call events over webhooks. See proof →
Indian languages
Hindi, Marathi, English and Hinglish in one call: language detection, per-language voices, script-aware guards. See proof →
Reliability and accuracy
Fallback chains for every provider, keyterm-prompted STT, voice-specific prompts that don't ramble or invent facts. See proof →
Shipping it
Docker and GitHub Actions to LiveKit Cloud, rate limits and abuse controls on public agents, latency budgets per stage. See proof →
Shipped
- InCBot — AI Video Interviewer · Shipped · Production @ InCruiter
- IncScreen — AI Phone & WhatsApp Screening · Shipped · Production @ InCruiter · built from scratch
- This site's AI voice clone · Shipped · Live on this page · open source
- 3 fixes sent to livekit/agents · in review · details
Stack I use in production
LiveKit Agents · WebRTC · Vapi · Twilio · Exotel · Telephony (PSTN) · STT / TTS · Deepgram · ElevenLabs · Silero VAD · Sarvam AI · Azure OpenAI · Groq · gpt-oss · LLM tool calling · Prompt engineering · LLM guardrails · Model fallback routing · Python · TypeScript · Node.js · React · Docker · Kubernetes · PostgreSQL · GitHub Actions · PyTorch · learning
Let's talk
Roles, collaborations or a voice agent that keeps breaking: hassankazinasik@gmail.com, or LinkedIn.