Software Engineer – Voice (Milan)

About the role

Indigo.ai is looking for a Software Engineer specialized in voice to own and evolve its real‑time voice runtime. The role is based in Milan and will work closely with the founders and the platform team to make phone‑based conversational AI feel human.

Key responsibilities

• Design, implement and maintain the end‑to‑end voice pipeline, from telephony edge audio ingress through streaming speech‑to‑text (STT), turn‑taking logic and text‑to‑speech (TTS) back to the caller.
• Optimize latency budgets and develop semantic end‑of‑turn detection, pre‑emptive generation on partial transcripts and filler/back‑channel strategies.
• Implement robust barge‑in and endpointing policies that handle noisy lines and preserve dialog state.
• Benchmark and A/B test STT and TTS providers on real G.711 8 kHz calls (Italian first), including wideband/HD voice where available.
• Build evaluation harnesses to turn subjective voice quality assessments into trusted metrics (WER, naturalness, numeric/code accuracy).
• Collaborate with the platform team to integrate voice runtime with the AI Agent “brain” and ensure compliance and knowledge guardrails.

Required profile

• Proven experience building real‑time voice or audio processing pipelines in production.
• Strong background with Node.js and TypeScript for high‑throughput services.
• Hands‑on knowledge of telephony media streams, codecs (G.711) and 8 kHz phone audio.
• Experience with streaming STT/TTS services and evaluating speech‑provider performance.
• Ability to own end‑to‑end architecture, latency optimisation and quality measurement.

Required skills

• Node.js
• TypeScript
• Streaming speech‑to‑text (STT)
• Streaming text‑to‑speech (TTS)
• Telephony media handling (G.711, wideband/HD)
• Real‑time audio processing
• Latency budgeting and optimisation
• A/B testing and benchmarking of speech providers
• Evaluation harness development