Principal / Staff Software Engineer

Nykaa; Bengaluru; Presencial; INR 2 500 000 - 6 000 000; Tempo integral

Resumo da oferta

Nykaa in Bengaluru is seeking an engineer to design and implement a streaming voice system with end-to-end response times under 800ms, progressing toward a seamless 500–600ms where users no longer notice AI interactions. You will enable human-like turn-taking, barge-in, and full-duplex audio with robust evaluation and observability.

Responsibilities include building the orchestration layer for tools and functions, applying RAG over customer knowledge, enforcing guardrails, and ensuring graceful

Qualificações

• Hands-on experience building real-time or streaming systems (voice, video, RTC, audio).
• Strong understanding of where milliseconds are spent and how to reduce latency.
• Familiarity with modern agent stacks: LLM orchestration, function calling, streaming pipelines, ASR/TTS, evaluation of non-deterministic systems.
• Have shipped real-time user interactions and achieved speed and reliability.

Responsabilidades

• Develop a streaming voice pipeline with end-to-end latency under ~800ms and a path toward 500–600ms.
• Implement human-like turn-taking, interruption handling, and full-duplex audio.
• Build an observability stack for latency budgets, transcription quality, hallucinations, and call traces.
• Create orchestration layer for tool/function calling, RAG over knowledge, guardrails, and graceful fallbacks.

Conhecimentos

Real-time systems

Latency optimization

LLM orchestration

Production experience

Descrição da oferta de emprego

Responsibilities

• A streaming, cascaded voice pipeline (ASR LLM TTS) with end-to-end response latency under ~800ms and a credible path toward the ~500-600ms range where callers stop noticing they're talking to AI.
• Turn-taking and barge-in that feel human - semantic endpointing, interruption handling, and full-duplex audio - not naive VAD silence timers.
• An evaluation and observability stack for voice: latency budgets per stage, transcription, hallucination and instruction-following evals, and replayable call traces.
• The orchestration layer: tool/function calling, RAG over customer knowledge, guardrails, and graceful fallback when a model or vendor degrades.

Requirements

• Deep, hands-on experience building real-time or streaming systems - voice, video, RTC, audio, or low-latency ML serving.
• You understand where milliseconds go.
• Working knowledge of the modern agent stack: LLM orchestration, function/tool calling, streaming token pipelines, ASR/TTS, and how to evaluate non-deterministic systems.
• You've shipped something users talked to (or otherwise interacted with in real time) and made it fast and reliable.