AI Research Scientist — Multimodal Systems

About AceSpeak

AceSpeak is an AI-powered interview and public-speaking coach for iOS and Android, developed by the Wistkey team. We are moving beyond the MVP stage and building a clearer, better-structured product development process.

The role

We are hiring an AI Research Scientist to turn applied research across speech, audio, video and behavioural signals into dependable product capabilities. You will design rigorous experiments, improve evaluation and calibration, prototype new coaching and scoring approaches, and work with engineers to move the strongest ideas into production.

What you will do

• Research multimodal methods for speech, audio, video and behavioural analysis
• Design experiments, benchmarks and evaluation frameworks
• Improve the reliability, calibration and explainability of scoring and feedback
• Prototype new AI coaching capabilities and validate them against measurable outcomes
• Work with engineering to productionise promising models and pipelines
• Investigate bias, failure modes, privacy risks and model limitations
• Document findings, trade-offs and recommendations clearly

What we are looking for

• 5+ years in applied AI research, machine learning or data science
• Strong Python skills and practical experience with PyTorch, TensorFlow or equivalent tools
• Experience in speech, audio, computer vision, NLP or multimodal systems
• Strong statistics, experimental design and model-evaluation skills
• A pragmatic, product-minded approach to research
• Clear written communication and collaborative working skills

Nice to have: a master's degree or PhD in a relevant field; publications, patents or evidence of shipped AI research; responsible-AI, fairness, privacy or human-centred evaluation experience.

AceSpeak candidate exercise

Shortlisted candidates will be asked to complete a short AceSpeak practice exercise and achieve a total AceSpeak score above 3,000.

1. Download the AceSpeak app.
2. Record a 90-second to 2-minute response to: “Describe a product or project you helped improve and what you learned.”
3. Practise and re-record until your total AceSpeak score is above 3,000.
4. Send us a screenshot showing your score and display name. Please do not send the recording.

We will provide shortlisted applicants with free access for this exercise; you do not need to purchase a subscription. The score is one part of a human-reviewed application process and is not used as the sole hiring criterion. If the app or scoring format is not accessible to you, or you need a reasonable adjustment, email [email protected] and we will arrange an equivalent alternative exercise.

Location and package

Full-time remote role for candidates based in the United Kingdom. £70,000–£95,000, depending on experience.

How to apply

Apply here: https://acespeak.uk/careers/ai-research-scientist-uk/

Questions or adjustment requests: [email protected]

Explore AceSpeak careers: https://acespeak.uk/careers/