US Mobile - Senior AI Engineer - Voice Biometrics & Conversational AI
Upload My Resume
Drop here or click to browse · Tap to choose · PDF, DOCX, DOC, RTF, TXT
Requirements
• Experience of at least 6+ years in AI/Machine Learning. • Strong portfolio experience specifically in Speech Processing or Audio Intelligence. • Proven expertise and practical work with speaker recognition techniques such as x-vectors/d-vectors, along with signal processing skills for biometric systems. • Deep understanding of the modern "Voice Stack," including knowledge on integrating ASR (Automatic Speech Recognition), LLMs (Large Language Models), and TTS (Text-to-Speech) into a cohesive workflow, with an emphasis on creating soph0. • - Experience in developing end-to-end low-latency pipelines for streaming audio to enable real-time interactions ("human-in-the-loop" or fully autonomous). • - Familiarity and experience working with Large Language Models (LLMs) as well as "Speech-to-Speech" architectures. • - Proficiency in performance engineering, specifically optimizing models for accuracy while minimizing false acceptance/rejection rates (FAR/FRR), along with ensuring sub-second response times ("Time to First Token"). • Expertise in PyTorch or TensorFlow and familiarity with speech libraries such as SpeechBrain.
Responsibilities
• Establish the Trust Layer by designing and deploying robust voice biometric systems for speaker verification and identification. • Develop advanced liveness detection and anti-cloning models to protect against deepfakes and synthetic audio attacks as part of Anti-Spoofing & Security measures. • Lead architectural efforts in transitioning toward autonomous Voice Agents by integrating Biometrics with ASR, TTS, NLU, and LLMs for pipeline innovation. • Build end-to-end low-latency pipelines that process streaming audio to enable fluid real-time interactions either "human-in-the-loop" or fully autonomous (Pipeline Innovation). • Explore and implement Large Language Models (LLMs) along with Speech-to-Speech architectures for giving voice agents contextual memory and reasoning capabilities. • Optimize models to achieve high accuracy while minimizing false acceptance/rejection rates, ensuring sub-second "Time to First Token" response times in Performance Engineering tasks.
Similar Jobs
No credit card. Takes 10 seconds.