Berkan Sesen

I co-founded tinie.ai with a focus on democratising access to AI for the populations that may benefit the most from it. Our first product is Recallify, which is a memory companion app that helps you capture memories in text, voice, or video, and allows you to summarize, search, and recall them by keyword or voice. It uses scientifically-based methods to monitor the difficulty of recalling information, providing insights into your recall ability over time, and generates personalized questions to help train and assess your memory using your own memories.

I co-founded A.I.M. research lab, which is a non-profit organisation that fosters collaboration among academics & industry professionals worldwide for publishing peer-reviewed, high-impact research papers.

I am part of the Techstars global startup network as a mentor. I currently mentor for the Techstars London and the ABN AMRO + Techstars Future of Finance accelerators.

I was the head of algorithmic execution at UBS Execution Hub, where I led the AI-driven algorithmic trading team consisting of quantitative analysts and developers. Our team owned the algorithmic trading / execution component of the business. We carried out cutting-edge Artificial Intelligence research in Bayesian Inference applied to market impact and transaction cost analysis for cash equities and cash bonds instruments.

I worked as a quantitative PM (and later senior PM), focusing on statistical and ML-based solutions that encompass the portfolio management, data science and trading domains. I built the strategies for and managed the ML-powered thematic funds:

• i. “JP Morgan Genetic Therapies” (AUM: $1.8B in Jan 2021) "JPGTAUA LX Equity"
• ii. “AU AM Future City” (AUM: $280M in Jan 2021). "AY31220A JP Equity"

As part of my quantitative research, I developed patent-pending ML/NLP applications to streamline stock selection for corporate action, ESG and thematic investing businesses. I also devised a transaction cost-aware portfolio construction framework for Equities investing, and supervised research on alternative data using scalable parallel computing infrastructures including PySpark and GPU computing (PyTorch, Tensorflow, etc.)

I was an honorary statistical inference researcher at the Fetal Medicine Unit at the St George's University Hospitals NHS Foundation Trust. We worked on designing an ML model to predict still births based on prenatal metrics.

I was a member of the algorithmic market making (AMM) team, trading CDS indices, and cash / government / inflation bonds. I also co-led the data analytics group for the global quantitative analysis department at Citigroup.
One of my early contributions was designing and implementing the recommendation algorithm for Citi’s global credit sales and trading platform: Vantage. Other projects included utilising Deep Reinforcement Learning for streaming adaptive bid/ask levels (market making), and research into Hidden Markov Models to identify market regimes based on various equity and credit signals. Our team also regularly carried out ad-hoc analyses such as client and trade clustering, RFQ pricing, trade P&L and client profitability analyses on demand.

I worked as a visiting researcher at the Biomedical Informatics department at Stanford University. My brief time here focused on clinical decision support, ontological reasoning (OWL), and probabilistic inference.

I was an honorary reseracher at the Oxford Radcliffe Trust. I attended the Lung cancer multi-disciplinary team (MDT) meetings in order to help design and inform my DPhil research: Lung Cancer Assistant.

I completed the Science Innovation Plus Programme at the Said Business School, which brought together MBA candidates and doctoral students from the life sciences departments. The programme included modules like Technology-Market-Organisation Analysis, Business and Strategy, Intellectual Property.

My DPhil focus was combining semantic reasoning (qualitative inference) and machine learning (quantitative inference) to help better inform lung cancer care decisions taken at multi-disciplinary team (MDT) meetings. It involved applying Bayesian Networks, (and toher various machine learning algorithms) on the large English Lung Cancer Database (LUCADA) to predict patient outcomes and recommend optimal treatments. This was complemented by lung cancer treatment guidelines encoded in a domain-specific (OWL) ontology to provide recommendations based on patient characteristics. If you are in any way excited, here is the full thesis (all 250 pages of it!).

I worked as a mathematical consultant and data analyst at the Laboratory for Industrial Mathematics (LIME) situated at the Maths department of Eindhoven Technical University. During my time here, I took part in three projects for: 1. European Space Agency, 2. Philips, and 3. ASML.

MSc at the University of Oxford, where I took key a very diverse set of key modules in biomedical engineering. I worked with Rene Banares-Alcantara for my thesis, which was later published. I graduated with a distinction.

During my undergradaute studies, I worked as a part-time business analyst at the Turkiye distributor of VW and all associated brands (Dogus Otomotiv).

BSc at Istanbul Technical University on chemical engineering. This is where I first got into programmanig, cybernetics and statistics. I graduated in the "high honour list" of my year.