Machine Learning Engineer / Voice-centric AI

We’re looking for an entrepreneurialSenior Machine Learning Engineerwith experience taking voice-centric AI systems (TTS, STT, LLM-driven dialog) from prototype to large-scale production. You’ll own the full ML lifecycle—research, data pipelines, training, evaluation, deployment, and ongoing optimization—powering sub-second, natural voice conversations at scale. This role is ideal for someone passionate aboutpushing the limits of conversational AI: creating highly optimized, domain-specific models that are faster, leaner, and more cost-efficient than general-purpose solutions. You’ll collaborate closely with product, infrastructure, and compliance teams, while setting the technical bar for model excellence and ML best practices. Required Skills & Experience • Experience: 7+ years building production ML systems, including 2+ years in speech or conversational AI. Proven track record deploying large-scale voice AI or LLM products. • Fine-tuning & compression (LoRA, QLoRA, quantization, pruning, distillation). • Speech (ASR: Whisper, NeMo, Kaldi; TTS: Tacotron, FastSpeech, VITS). • LLMs & dialogue (GPT-class, RAG, LangGraph, LangChain, MCP). • Strong in Python; bonus for TypeScript/Node/Java. • Infra & Ops (Kubernetes, Helm, Terraform, MLflow/SageMaker). • Data systems (Kafka, Redis, Postgres, Snowflake). • Streaming protocols (gRPC, WebSockets, HTTP/2, WebRTC). • Security & compliance (HIPAA, SOC2, HITRUST). • Product-oriented, entrepreneurial, strong problem solver, effective communicator, and technical leader.Desired Skills & Experience • Optimize & Fine-Tune Models: Apply LoRA, QLoRA, RLHF, and other parameter-efficient techniques. Use quantization, pruning, and distillation to shrink models while preserving quality. • Build End-to-End Pipelines: Design STT, TTS, and LLM systems achieving <800 ms latency and thousands of concurrent sessions. • Scale Inference: Optimize serving on Kubernetes/EKS with dynamic batching, speculative decoding, and streaming protocols. • Advance Dialogue Management: Extend LangGraph/LangChain flows and MCP schemas for complex multi-turn conversations. • Data & Evaluation: Develop pipelines for conversational logs (Kafka ? Snowflake/S3) and create frameworks to measure ASR accuracy, TTS quality, and task completion. • Lead & Mentor: Define ML best practices, champion model CICD and monitoring, and mentor teammates on ML Ops, speech processing, and prompt engineering. • Innovate & Research: Run POCs with cutting-edge models (e.g., Whisper-v3, Bark) and stay ahead of the latest in speech + LLM research. • Ensure Reliability & Compliance: Implement HIPAA-grade security, PHI safeguards, and robust fallback strategies.The Offer • Bonus eligibleYou will receive the following benefits: • Medical, Dental, and Vision Insurance • Vacation Time • Stock OptionsApplicants must be currently authorized to work in the US on a full-time basis now and in the future. #LI-EM1