Meet Our New Open-Source AI Fellows
Senior AI Innovation Manager
18 February 2026
i.AI are welcoming seven world-class AI researchers through the Open-Source AI Fellowship to demonstrate the real-world potential of open-source AI. Over the next year, fellows will tackle critical public sector challenges spanning transport, national security, education, health, and policing, building AI solutions that are transparent, secure, and built for the public good. Learn more about our first cohort of fellows in this blog.
Bringing AI Expertise Into Government
We're excited to introduce seven exceptional fellows joining the Open-Source AI Fellowship. Backed by Meta and delivered through the Alan Turing Institute, this initiative brings leading AI researchers and practitioners into government to tackle some of the biggest challenges facing public services.
The Fellowship is all about demonstrating what open-source AI can really do. Open-source models offer transparency, cost-effectiveness at scale, and freedom from vendor lock-in, but their use in government has been limited so far. Our fellows will spend the next year showing how these tools can work in practice, building solutions that are interpretable, secure, and designed with the public interest at heart.
What Will the Fellows Be Working On?
The Fellows will focus on work spanning key frontline areas.
- In transport, Fellows will use AI to develop computer vision models for infrastructure inspection, helping councils prioritise repairs roads more intelligently with the same resources.
- In national security, specialists will develop state‑of‑the‑art air‑gapped language models. This will enable defence and national security professionals to use modern AI safely, securely and entirely offline to analyse sensitive information and support critical decision-making.
- In education, Fellows will support initiatives to establish a new benchmark to assess pedagogy and safety of AI models against UK standards.
Further projects are also being scoped across health, policing and GOV.UK chat, extending the impact of the programme across the heart of public service delivery.
Armin Mustafa
Background, research interests, and key accomplishments:
I am an Associate Professor in Computer Vision and AI at the University of Surrey's CVSSP. I am also the Co-Founder and Chief Scientist at SAIReco Ltd. With over a decade of experience in 3D/4D Vision, my research focuses on scene understanding, multi-modal AI, and immersive technologies like AR/VR. I was formerly a Royal Academy of Engineering Research Fellow and Lead Engineer at Samsung Research, bridging the gap between academic innovation and industrial application. I have published over 50 papers in premier venues, and as a Tech Women 100 award winner, I am dedicated to advancing human-like machine perception.
What has motivated you to join the Fellowship?
Joining the fellowship is driven by my commitment to translating cutting-edge research into public benefit. As an expert in 4D vision, video understanding, and multi-modal AI, I am motivated to leverage open-source models to build transparent, sovereign capabilities for the UK. This fellowship offers a unique opportunity to apply my academic insights to critical societal challenges. By bridging the gap between high-level academia and government policy, I aim to foster a responsible AI ecosystem that ensures technology serves the public interest first.
Angus Williams
Background, research interests, and key accomplishments:
I am a computer scientist with the Alan Turing Institute, doing applied research on evaluating LLMs and ML for safety. My work centers around applying state-of-the-art NLP methods and models to real-world use cases and data streams. I have published peer-reviewed work on training generalisable transformer models and AI safety, and built production ML systems at scale, including evaluation infrastructure, agentic AI pipelines, and products processing millions of data points daily. I work closely with government and regulators to advise on the impacts of emerging technology.
What has motivated you to join the Fellowship?
AI has huge potential, but realising it depends on doing it well: thoughtful design, understanding what's actually needed, building systems that work in practice. That's what draws me to i.AI, a community that cares about getting this right, in a context where it really matters. Government is one of the highest-impact places to be building AI systems, in terms of scale, stakes, and the potential to improve how public services work. I am excited to be part of a team shaping how that happens, and to learn a lot in the process.
Frank Soboczenski
Background, research interests, and key accomplishments:
I am an Assistant Professor in AI/ML at the University of York and an affiliate scientist at King's College London. My research focuses on deep learning, explainable AI, Transformer models, and quantum machine learning, with applications in space science and healthcare. I am a member of ELLIS, a STEM scientist for NASA/NOAA's GLOBE programme, Huggingface BigScience and actively contribute to multiple NASA groups and international consortia. I have collaborated with GCHQ, Airbus, Google, INRIA, and many others. I received multiple international recognitions such as the NASA TechLeap Prize and recently contributed to the first federated causal-inference platform on the ISS.
What has motivated you to join the Fellowship?
I am motivated by the opportunity to contribute my expertise in cutting-edge artificial intelligence to strengthen and modernise government operations. The Fellowship offers a unique platform to apply advanced AI methods to improve efficiency, transparency, and evidence-based decision-making across public services. By working closely with and across government departments, I aim to help translate state-of-the-art research into practical solutions that deliver real societal impact, enhance government operations, and ensure that emerging technologies are deployed responsibly, securely, and for the benefit of the public.
Shan Luo
Background, research interests, and key accomplishments:
I am a Reader in Robotics and AI and Director of the Robot Perception Lab at King's College London. My research focuses on AI for multimodal perception, representation learning, and autonomous decision-making, with robotics serving as a testbed for developing and validating these methods. I work on learning from visual–tactile data, foundation models for perception, and data-efficient autonomy, with applications in manufacturing, laboratory automation, and healthcare. I have led and contributed to major UKRI-funded projects, hold both the EPSRC Open Fellowship and New Investigator Award, and serve as an Associate Editor for IEEE Transactions on Robotics.
What has motivated you to join the Fellowship?
I am motivated to join the Open Source AI Fellowship because it offers a rare opportunity to align advanced AI research with real civil service challenges, while ensuring transparency, reproducibility, and public benefit through open-source development. The Fellowship's emphasis on deploying AI responsibly within government strongly resonates with my interest in translating research into trusted, scalable systems that support policy, public services, and scientific discovery. I am particularly excited by the opportunity to co-develop open models and tools with delivery partners, ensuring they are robust, interpretable, and accessible, and that their impact extends beyond individual projects to benefit wider society.
Chris Willcocks
Background, research interests, and key accomplishments:
My background is in both theoretical deep learning and large-scale applied generative modelling, having created Durham's deep learning and reinforcement learning courses over the past decade. Key contributions include benchmark evals widely used by frontier AI labs, several widely cited advances to generative diffusion models, and self-regulated sampling strategies in frontier LLMs. I am particularly interested in multimodal AI. Current systems offer bespoke solutions for text or images, but real-world data is far more diverse. Future AI must generalise across modalities at scale. I am also fascinated by neural scaling laws for decentralised AI, and how to evaluate and improve agentic systems working together.
What has motivated you to join the Fellowship?
As an academic, getting close to real-world high-impact problems affecting thousands of people can be challenging. My work with UK national policing has shown me the value of deployed AI systems at scale. This fellowship offers an exceptional opportunity to apply research directly at the heart of government, while being part of a broader network of frontier AI labs collaborating with the Turing Institute on large-scale Open Source AI.
Ameet Patil
Background, research interests, and key accomplishments:
I am a seasoned entrepreneur and technologist with 20+ years of experience founding, scaling, and leading companies across the UK, India, UAE, and Africa. I hold a PhD in Computer Science from the University of York, combining deep technical expertise with strong business acumen to build scalable, enterprise-grade solutions. I have successfully delivered platforms serving over 1M users and processing 2M+ documents daily, navigating complex regulatory and cultural environments. I am the Founder & CEO of Ecobillz, a GDPR, ISO 27001, and SOC 2 Type II compliant Digital Automation Platform, working with senior leaders to drive operational excellence and data-driven insights at a global scale while mentoring high-performance teams.
What has motivated you to join the Fellowship?
The Fellowship represents a unique opportunity to apply my experience in building scalable, technology-led solutions to challenges that truly matter at a national level. Throughout my career, I have been motivated by the ambition to create systems that operate at scale and deliver meaningful impact for people. This Fellowship aligns closely with my aspiration to contribute to initiatives that strengthen the connection between government and society by designing inclusive, reliable, and efficient digital platforms. By leveraging my background in enterprise technology and large-scale systems, I hope to support the development of AI solutions that improve accessibility, transparency, and service delivery for people across the country.
Mingfei Sun
Background, research interests, and key accomplishments:
I am a Lecturer (Assistant Professor) in the Department of Computer Science at The University of Manchester, working on artificial intelligence with a focus on deep reinforcement learning and generative models. Previously, I was a Researcher at Microsoft Research Cambridge. My research develops scalable policy optimisation and generative methods grounded in optimisation theory, with work recognised by a Best Paper Award at AAMAS. I have secured competitive funding from EPSRC, UKRI, Innovate UK, and industry partners, supporting the development of robust, open, and deployable AI systems.
What has motivated you to join the Fellowship?
I am motivated to join the Open-Source AI Fellowship because it enables the translation of advanced AI research into open, transparent tools that benefit the public sector. My research focuses on building scalable and reliable Reinforcement Learning algorithms, and I am eager to apply this expertise to real-world challenges in government and public services. The fellowship's strong emphasis on open-source development, collaboration, and responsible AI closely aligns with my commitment to reproducible research and public-interest AI.