Speechmatics, Engineering
Senior Machine Learning Engineer at Speechmatics
- Location
- London, United Kingdom
- Compensation
- £95k + Bonus + Equity
Speechmatics is looking for Senior Machine Learning Engineers to join their Modelling Team in London or Cambridge. As a world leader in Voice AI, Speechmatics is on a mission to "Understand Every Voice" by shipping the most accurate speech recognition models in existence.
You’ll work on the latest transformer architectures and optimize inference for production using dynamic batching and flash attention. This is a rare chance to take a senior seat on a team training frontier models while enjoying £95,000 base pay, equity, and "Focus Fridays." If you have experience in distributed training and a passion for R&D, this is your opportunity to define the future of audio AI.
Role overview
Join the Modelling Team at Speechmatics to bridge the gap between frontier research and production-grade AI. You will lead complex R&D projects, training diverse models and optimizing inference for global scale. This role offers the chance to define the future of speech recognition while mentoring a talented, diverse engineering group.
About Speechmatics
Software
Speechmatics is a technology company specializing in artificial intelligence-driven speech recognition and voice AI infrastructure. It provides automatic speech recognition (ASR), speech-to-text (STT), text-to-speech (TTS), translation, summarization, sentiment analysis, and topic detection services, supporting over 50 languages and dialects with high accuracy across accents, noisy environments, and multi-speaker scenarios. The company serves enterprises, developers, and partners in industries such as media & broadcast, contact centers, healthcare, education, legal, finance, and AI infrastructure, enabling real-time and batch transcription for applications like captioning, compliance, and voice agents.[1][5]
What you will do
- Design and train diverse machine learning models, including large self-supervised architectures, and take them from R&D into production.
- Optimize inference performance using advanced techniques like dynamic batching, flash attention, speculative decoding, and KV-caching to achieve real latency wins.
- Lead high-impact projects across departments, mentoring teammates and ensuring the team maintains the highest standards for model serving and engineering.
Who this is a fit for
- Possesses deep, hands-on experience in distributed training and optimizing modern transformer architectures like GQA for production environments.
- Brings prior expertise in speech or audio modeling, or demonstrates a strong background in LLMs with a desire to transition into voice AI.
- Thrives in collaborative R&D settings, with a meticulous eye for detail and a proven track record of shipping models that deliver cost and performance gains.
Why this role is remarkable
- Work at the absolute frontier of Voice AI, training large self-supervised models that power the world’s most accurate speech recognition technology.
- Join a world leader in the field that blends high-impact R&D with rigorous software engineering practices to ship real-world AI solutions.
- Enjoy a high-trust culture with “Focus Fridays” for deep work and “Together Tuesdays” for collaboration, supported by a comprehensive benefits package.
How Jack & Jill work together

What happens next?
Jack’s an AI agent for job searching and career coaching. He works for you.
Jill is the AI agent for hiring, working for the company. She recruits from Jack’s network.
If your profile’s a match and Speechmatics wants to meet, Jill will make the intro. In the meantime, Jack will send you excellent alternatives.