About
Valerio Ciotti
Data scientist · network scientist · tinkerer
I am a data science professional with more than eight years of experience in banking and fintech, in large and small organisations. I work in Python, big data and machine learning, and I like solving business problems with data, in a team, by trying ideas early.
Today I lead a data science team in retail banking in Amsterdam, working on churn, propensity to buy and recommender systems. Before that I was a data scientist at Barclays in London, and lead data scientist at FXCintel.
I received my PhD from the School of Mathematical Sciences at Queen Mary University of London, in the Complex Systems and Networks group of Prof. Vito Latora. Networks still shape how I look at almost everything.
Machine learningNetwork analysisPythonReinforcement learningComputer visionLeadershipRaspberry Pi
The way here · open any stop for the details
-
Leads a team of 10 data scientists in the retail banking centre of excellence.
- Appointed to lead the data science team of the retail banking centre of excellence, with data scientists from junior to senior.
- Builds analytical models for eCommerce and engagement, customer experience, digital channels, customer interaction and the contact centre.
- Leads a long-running, bank-wide project to choose the next best action for each customer, in line with the bank's strategy.
- With the team, delivered more than five product use cases. Model-driven audience selection raised conversions by 30–70%, a recommender raised newsletter engagement by 80%, and a model now sorts open-text reviews into a taxonomy automatically.
-
A 20-year study of founder-led, VC-backed tech IPOs against the market, for the CEO of a fund.
- Evaluated an investment strategy for a fund that invests in founder-led companies.
- Studied the market performance of venture-backed tech companies that went public with their founder still as CEO.
- Tested whether they beat the market composite index over twenty years, and delivered the results as a written report to the fund's CEO.
-
Scientific lead of the team; NPS drivers, churn and propensity models, a recommender.
- Scientific lead for the data science team, guiding colleagues towards solutions for their use cases.
- Turned unstructured click data into successful and unsuccessful customer journeys, the groundwork for a next-best-action system.
- Found the drivers of the bank's net promoter score, with a method written up for a scientific paper.
- Product owner for churn prevention and propensity-to-buy models, and helped build a collaborative-filtering recommender for newsletter articles.
- Member of the Data Science Review Board, setting good practice for teams across the bank.
-
LondHeNet, performance indicators and placement optimisation for Helium hotspots in London.
- Designed and ran the data science behind LondHeNet, a research project sponsored by the Decentralized Wireless Alliance.
- Defined performance indicators for the Helium hotspots already active in London and predicted the performance and rewards of new ones.
- Built a pipeline of models and simulations to choose where new hotspots should go, and worked with the app developer to serve the results in real time in an Android app.
-
Forecasting UK spending during COVID, start-up discovery with Barclays Ventures, the Data Bites talks.
- Technical lead on forecasting total UK spending from transaction data and customer behaviour, to follow the economy's recovery from the COVID crisis.
- Proposed and led a project to find promising early-stage start-ups with Barclays Ventures and the Eagle Labs accelerator.
- Technical lead on two confidential projects with Barclays Ventures, on transaction insights and on cyber-security.
- Part of the winning team of a bank-wide, Kaggle-style competition on credit default.
- Founded the Data Bites talk series, member of the Innovation Hub, and mentor in the Women in Machine Learning programme.
-
Dynamic FX price optimisation, lead generation from unstructured data, crawlers in production.
- Built and led a dynamic price optimisation system for foreign exchange providers, combining models with customer behaviour. It went into production in the company's client portal.
- Found and sized import and export businesses for FX providers from unstructured data such as company accounts, with gradient-boosted models.
- Wrote web crawlers and scrapers, automated with Jenkins and Docker, to enrich the company's data.
- Presented results to technical and non-technical clients, and supervised junior data scientists and analysts.
-
School of Mathematical Sciences, Complex Systems and Networks group of Prof. Vito Latora.
- Thesis on positive and negative connections and homophily in complex networks.
- Papers on degree correlations in signed social networks, missing links in citation networks, and predicting success in the worldwide start-up network.
- The reason almost everything on this site ends up drawn as a graph.