Freelance AI/ML Engineer

• • Please note: We welcome hybrid working, but applicants should be based in the UK and able to attend some in-person working sessions at our Brighton studio • • Can LLMs help promote experiences of connection and belonging within local communities? We’re looking for a skilled AI/ML engineer to join a creative research project with award winning interactive artists Blast Theory. Blast Theory are seeking an AI/ML Engineer to join in the development of a new interactive artwork based on the acclaimed project Rider Spoke. First created in 2007, Rider Spoke is an immersive experience where participants explore their city guided by a mobile app, discovering hidden places and personal messages left by those who came before them. The project has toured internationally, creating an archive of tens of thousands of recordings from participants. Rider Spoke invites deep reflection and has prompted an unexpected sense of connection among those who have taken part. 18 years on, Blast Theory are seeking to explore how AI might be used to distribute the project more widely - using securely-hosted AI-enabled services to facilitate the participant experience, and to assist with back-end processes such as transcribing and analysing recordings. ABOUT THE POST You will be working alongside the Lead Artist and app developer, Nick Tandavanitj, in a collaborative, creative, and research-driven environment to develop AI-enabled services that integrate with the Rider Spoke app. These services may include: • Transcribing speech • Analysing and classifying recording transcripts • Dynamically generating follow-up prompts for participants At the core of the project is a commitment to privacy. Therefore, you should have experience architecting and deploying services using self-hosted models and be able to ensure that user data never leaves trusted environments. You’ll collaborate with the Lead Artist on discussing appropriate architectures and technologies for these services and develop secure APIs for these to integrate with the app’s existing technology stack (PostgresQL, Node/TypeScript and React). As part of this work, you’ll be asked to share your knowledge: mentoring the Lead Artist to co-develop AI-enabled services with you, maintaining documentation for the services you create, and passing on technical skills in integrating AI/ML pipelines with web applications. Lastly, you’ll engage with feedback from user tests and contribute to discussions about the potential of AI as it is explored in this setting; reflecting on any ethical dilemmas that arise from the project. ABOUT BLAST THEORY “The variety of experience at Blast Theory is unmatched. We've built systems to make phone boxes ring in Hull, made Europe's first interactive feature film, participated in parades in the USA, and worked on a very unconventional life-coaching app. It's a supportive and thrilling environment to develop in and work on creative solutions to novel problems of our own making.” - Michael Kane, Software Lead at Blast Theory (2017-22) Blast Theory make interactive art to explore social and political questions. The group’s work places the public at the centre of unusual and sometimes unsettling experiences, to create new perspectives and open up the possibility of change. Led by Matt Adams and Nick Tandavanitj, the group draw on popular culture and new technologies to make performances, games, films, apps and installations. Blast Theory have shown work at the Venice Biennale, Tribeca Film Festival, ICC in Tokyo, Hebbel am Ufer in Berlin, the Barbican and Tate Britain. Commissioners include Channel 4, Sundance Film Festival and the Royal Opera House. ABOUT NICK TANDAVANITJ Nick is an artist in Blast Theory and is undertaking a practice-based PhD at the University of Nottingham exploring ways of reimagining Blast Theory’s work for new contexts and audiences. He has experience collaborating on digital projects and is seeking to build practical and technical skills around integrating AI and LLMs in creative applications. ABOUT BRAID The project is funded by BRAID (Bridging Responsible AI Divides), a UK-wide programme dedicated to integrating Arts and Humanities research more fully into the Responsible AI ecosystem, as well as bridging the divides between academic, industry, policy and regulatory work on responsible AI. You can read more about BRAID at braiduk.org. RESPONSIBILITIES • To work with the Lead Artist to develop technical proposals for integrating AI-enabled services including speech-to-text, text classification, and text generation with the Rider Spoke application. • To set up secure environments for development, testing, training and deployment of these services. • To develop, test and deploy these services, providing robust and secure APIs for the Rider Spoke application to use them. • To establish processes for monitoring, testing and troubleshooting these services to support their long term use. • To mentor the Lead Artist, sharing skills and knowledge through discussions and presentations of your work, sharing or assigning development tasks with the Lead Artist, taking part in code reviews and by maintaining up-to-date documentation of your work. • To engage with feedback from internal tests and user tests to modify these services in discussion with the Lead Artist. • To contribute to creative and reflective discussions about the potential and risks of AI that emerge from its application in this setting. SKILLS AND COMPETENCIES Essential • Good understanding of current technologies and approaches for integrating AI-enabled services with web based applications. • Strong Python development skills for ML/NLP workflows, with experience of running NLP models securely on cloud servers (Docker, Digital Ocean) • Experience running local speech-to-text models (Whisper.cpp, Vosk) • Experience with training, deploying and testing language models for classification and generation tasks. • Experience using prompt engineering for dynamic, context-aware text generation • Proficiency with Docker for environment setup and deployment • Good knowledge of data privacy and implementing security practices in this setting. • Working knowledge of Git for shared development and version control. • Strong communication skills, with the ability to explain technical concepts and produce documentation to a good standard • Willingness to mentor a technically skilled developer and share knowledge of current best practices • Aptitude for working in research environments and curiosity for working with creative, open-ended processes Desirable • Prior involvement in interactive art, creative coding, or research-led projects • Familiarity with embedding models and vector search • Knowledge of Typescript, full-stack Javascript applications and PostgresQL • Awareness of AI ethics and critical perspectives on generative AI KEY INFO AND TIMELINE Fee: £350/day Engagement: • Trial day (Jun 2025 ) • Development, Phase 1 (15 days, Jul-Sep 2025) • Development, Phase 2 (up to 10 days, Oct-Nov 2025) Location: We welcome hybrid working, but applicants should be based in the UK and able to attend some in-person working sessions at our Brighton studio. TRIAL (1 DAY) One day, in-person at Blast Theory’s studio to get to know the project, kickstart work and assess fit. You will: • Learn about Rider Spoke: the participants’ experience, the technology stack in the current app and the archive of recordings that accompany the project. • Discuss potential NLP tasks for AI-enabled services and reflect on privacy, security and ethical considerations for these tasks. • Advise on architecture and specific models, AI technologies and deployments that may be considered. • Work with the Lead Artist to set up a simple proof-of-concept for one of the NLP tasks identified (e.g. sentiment classification) • Work with Blast Theory’s Project Manager and Lead Artist to sketch out a work plan for development. Deliverables • A summary of AI-enabled services discussed, including technical recommendations • A proof-of-concept demonstration of an agreed NLP task • A draft work plan for development DEVELOPMENT, PHASES 1 & 2 This work is part time with specific dates to be agreed. It may be completed entirely in person at Blast Theory’s studio, or remotely with key meetings and milestones taking place in person. HOW TO APPLY Download the full job description here. To apply, fill out this form. Deadline: Monday 26 May 2025 at 5pm The form will ask you for the following: • A short cover letter (max 1 page) • Your CV (or link to a portfolio/GitHub) • Any relevant examples of work We’ll be reviewing applications as they come in and may fill the role before the deadline, so don’t wait to apply! If you have any questions please email Kizzie at kizzie@blasttheory.co.uk EQUAL OPPORTUNITIES We’re an equal opportunities employer and actively encourage applications from people underrepresented in the arts and tech sector, including people of colour, people with disabilities or mental health conditions (including ‘invisible’ disabilities such as dyslexia, ADHD or ASD), people who identify as LGBTQIA+ and people from lower socioeconomic backgrounds.