Senior Machine Learning Engineer

About the role

Retell AI is using first-principles thinking to reimagine the call center with cutting-edge voice AI. Thousands of companies now use Retell's AI voice agents to handle sales, support, and logistics calls that once required large teams of human agents. This Senior Machine Learning Engineer role is a hands-on, high-ownership position where you will own the end-to-end lifecycle of production voice AI models. You will fine-tune large language models and audio models, evaluate them with rigorous benchmarks and human feedback, and deploy them into latency-sensitive, high-traffic systems that power real-time phone conversations. You will define and drive our ML strategy alongside the founding team, shaping how intelligent AI "workers" act as frontline agents, QA analysts, and managers. If you are excited by hard technical challenges, fast iteration, and building voice AI at scale from the ground up, this role gives you the opportunity to have a real impact on one of the fastest-growing companies in the space.

Key facts

What you'll do

• Train & Tune Models - Fine-tune LLMs and audio models to maximize speed, accuracy, and production-readiness for real-time AI voice experiences.
• Benchmark & Evaluate - Build datasets, define rigorous metrics, and measure model performance across high-impact voice AI tasks to guide development and prioritize improvements.
• Deploy to Production - Work closely with engineering to ship models, monitor them in the wild, and ensure they stay fast, reliable, and accurate at scale under real-world traffic conditions.
• Run Human Evaluations - Build scalable pipelines to collect structured human feedback, benchmark subjective quality, and translate findings into concrete model iterations.
• Level Up Infrastructure - Design and maintain the ML infrastructure needed for fast experimentation, robust training workflows, and continuous deployment of voice models.
• Own Model Performance End-to-End - Drive improvements from training data through deployment monitoring, diagnosing regressions, and implementing fixes that preserve latency and reliability targets.
• Shape Technical Roadmap - Collaborate with product and engineering leadership to define evaluation benchmarks, experiment priorities, and long-term model strategies for the voice platform.
• Mentor and Collaborate - Work cross-functionally with data scientists, engineers, and product teams to align model capabilities with business goals and customer needs.
• Optimize for Real-Time Constraints - Ensure models meet strict latency and throughput requirements for millions of concurrent phone calls while maintaining high voice quality and accuracy.
• Contribute to Best Practices - Document experiments, training procedures, and evaluation results to create reusable knowledge and accelerate future hiring and onboarding.

Requirements

• ML Engineer with Real-World Experience - You have trained and shipped models in production, with bonus experience in LLMs or audio models in voice AI or related domains.
• Fluent in Modern ML Stack - You are deeply proficient in Python, PyTorch, and today's ML tools, covering training pipelines, evaluation benchmarks, and deployment workflows.
• Execution-Oriented Mindset - You move fast, take ownership, and focus on solving real problems with pragmatic solutions rather than seeking perfect ones upfront.
• Startup-Ready Resilience - You are adaptable, resilient, and energized by ambiguity and fast-changing priorities in a high-growth environment.
• Clear Communicator & Team Player - You communicate clearly across functions and collaborate effectively with engineers, product managers, and leadership to push decisions forward.
• Strong Analytical Skills - You design experiments, analyze results, and use data to drive model improvements and infrastructure decisions.
• Commitment to Quality - You build rigorous evaluation practices, including human evaluations, to ensure model outputs meet real-world standards.
• Infrastructure Awareness - You understand the constraints of production systems and design models with latency, reliability, and scalability in mind.

Nice to have

• Experience with voice AI, speech recognition, or audio generation pipelines.
• Familiarity with large language model fine-tuning frameworks and parameter-efficient tuning methods.
• Contributions to open-source ML projects or presence in relevant technical communities.
• Experience building and maintaining human evaluation platforms or feedback loops for AI systems.
• Knowledge of deployment tooling for ML such as model serving frameworks, monitoring systems, and experiment management tools.

Practical notes

• Cash compensation is specified as $225,000 - $325,000 base salary.
• Equity offers are part of the total compensation package.
• Location is listed as Redwood City, California, US, within the San Francisco Bay Area.
• The company is open to sponsoring multiple work authorization types including H1B/H-1B, TN, L-1, E-3, F-1 (OPT/CPT), and O-1 visas.
• No specific working hours are stated, implying standard full-time expectations in a startup context.
• No explicit travel requirements are listed, though the role may involve occasional collaboration or onsite presence as needed.
• No application deadline is communicated, indicating continuous review of candidates while the role remains open.