Berkan Sesen
18+
years of experience
years of experience
Applied AI Adviser — Oxford DPhil, ex-J.P. Morgan, UBS, Citi
Berkan has worked on AI in high-stakes settings since 2009, starting with an Oxford DPhil in AI and Medical Informatics, where he built clinical decision support systems that predicted patient outcomes and recommended treatment, published in leading medical and interdisciplinary journals. He has published on AI continuously since, most recently on the evaluation of retrieval-augmented LLMs in clinical use, and writes for industry on AI transformation and clinical AI.
Between then and now he spent a decade running AI-driven products where the outcome was measured in money. At J.P. Morgan Asset Management he took two ML-powered thematic funds to over $2bn in combined AUM, the underlying methodology becoming a US patent application. He then headed Algorithmic Execution and AI at UBS, owning the algorithms and the product for the bank's in-house multi-asset execution platform and running the team that built it. Before that, at Citi, he co-led the global data analytics group in Markets Quantitative Analysis.
In 2024 he co-founded Recallify, a memory and learning app for people with ADHD, brain injury and cognitive fatigue, built alongside an NHS neuropsychologist as clinical co-founder and now live on both app stores. He runs sesen.ai, an AI and machine learning consultancy, and has mentored AI startups at Techstars since 2019 and at AI Forge.
Half the job is working out which parts of the problem actually need AI. The other half is building those parts properly.
Good
match for
Founders making decisions about AI in their product: what to build, whether to build it at all, and what it takes to hold up with real users, real costs and real regulation. Particularly strong fit in health, clinical, finance or other regulated and sensitive-data settings.
Berkan Sesen's industry experience
Health tech & clinical AI
Financial services & asset management
AI transformation & enablement
Nonprofit & international development
When to bring us in
Typical decisions
The critical technology decisions leaders face when delivery, investment or product direction is at risk.We have an AI prototype that works, what does it take to get it to something scalable, robust and production-ready?
We're being told the model performs well. How do we know that number is real and not an artefact of how it was tested?
We're moving from research into a regulated or high-stakes environment. How do we validate it properly?
What's the right AI strategy and roadmap for a platform and team that already exist?
Examples
Led two ML-powered thematic funds at J.P. Morgan Asset Management to $2bn+ in combined AUM, methodology filed as a US patent application; co-developed News Filter, winner of the Banking Technology Awards UK and the American Financial Technology Awards
Product and algorithm owner for UBS's in-house multi-asset execution platform across equities and bonds, running the team of quantitative researchers and engineers who built it
Co-led the global data analytics group in Markets Quantitative Analysis at Citi, building AI into algorithmic trading and market-making across bond markets
Built Recallify's full stack, from on-device speech-to-text and retrieval to the app-store release pipeline; live on both stores, UKABIF Innovation Award, NIHR and Innovate UK funded
Publishing on AI since 2009 across clinical decision support, quantitative finance and generative models: peer-reviewed journals, two US patent applications, and book chapters with Wiley Finance and CIO
Articles
Peer-reviewed publications:
Google Scholar (opens in a new tab)Grand Rounds Availability
Coming soon