ML
ML Interview Roadmap
← Study plan·System Design

Day 119 of 133

Design real-time voice agent (ASR → LLM → TTS) + DSA review

Turn-taking; barge-in; first-audio-out latency budget.

DSA · NeetCode Heap / Priority Queue

  • Find Median From Data StreamDSA · Heap / Priority Queue

    Interview questions to prep

    1. Walk through the two-heaps trick (max-heap left, min-heap right). What invariant ties them?
    2. What's the space cost over a long-running stream, and how would you bound it (windowed median)?

ML System Design · Voice agent

  • Design a real-time voice agent (Sesame/ChatGPT voice)ML System DesignSesame

    Interview questions to prep

    1. Walk me through designing a real-time voice agent (ASR → LLM → TTS).
    2. How do you handle interruption, turn-taking, and barge-in?
    3. What's your latency budget end-to-end (first audio out)?

References & further reading