Senior Machine Learning Engineer - Automatic Speech Recognition (ASR) in London

Cresta is on a mission to turn every customer conversation into a competitive advantage by unlocking the true potential of the contact center. Our platform combines the best of AI and human intelligence to help contact centers discover customer insights and behavioural best practices, automate conversations and inefficient processes, and empower every team member to work smarter and faster.

About The Role

At Cresta, we are dedicated to building state-of-the-art Machine Learning systems that power real-time, intelligent customer interactions. Our team develops models and platforms that process large-scale, multimodal data—especially speech and text—to extract meaning, improve quality, and deliver actionable insights at scale. A key focus of this role is advancing model evaluation, measurement, and quality improvements, with particular emphasis on Automatic Speech Recognition (ASR) and downstream NLP systems. You will design rigorous evaluation frameworks, define quality metrics, and drive systematic improvements to model accuracy, robustness, and reliability.

Responsibilities

• Design, implement, and maintain evaluation frameworks to measure model accuracy, robustness, latency, and real-world performance across ASR and NLP systems.
• Lead ASR quality improvement efforts, including error analysis, dataset curation, metric definition (e.g., WER and task-specific metrics), and model iteration.
• Analyze large-scale speech and text data to identify failure modes and drive targeted model and data improvements.
• Develop, train, and deploy machine learning models for speech recognition and downstream tasks such as classification, entity recognition, information extraction, and structured insight generation.
• Partner with applied research to translate experimental improvements into production-ready systems.
• Collaborate with product managers, platform engineers, and UX teams to align model quality metrics with customer and business goals.
• Optimize ML pipelines and evaluation workflows to operate efficiently and reliably at scale.
• Establish best practices for model validation, offline/online evaluation, and continuous quality monitoring in production.

Qualifications We Value

• Master's or Ph.D. in Computer Science, Machine Learning, AI, or a related field.
• 5+ years of hands-on experience building, evaluating, and deploying ML models in production.
• Strong background in speech recognition (ASR), speech processing, or closely related domains.
• Deep experience with model evaluation, benchmarking, and error analysis for ML systems.
• Proficiency with ML frameworks and libraries (e.g., PyTorch, TensorFlow, Hugging Face).
• Solid understanding of modern ML techniques, including transformer-based models and large-scale training.
• Experience building data pipelines and tooling for large-scale experimentation and quality analysis.
• Strong passion for improving real-world AI system quality, with a track record of delivering measurable, production-grade improvements.