Romain Caristan

About Me

🚀 Passionate about new technologies and AI, my focus is on solving complex problems and driving informed decision-making through the power of data-driven approaches.

📊 With my skills in programming and statistical analysis, I embark on exciting journeys with large datasets. I uncover hidden gems of information, extracting meaningful insights that pave the way for predictive models.

🤖 I firmly believe in the importance of responsible AI practices, ensuring fairness and transparency in all solutions I develop. Together, let's create a future where cutting-edge AI technologies make a positive impact on society.

🏢 Currently, I am working on innovation projects for the Cross-Asset pricing platform in the UBS Investment Bank, pushing the boundaries of what's possible.

🇫🇷🇨🇭 Originally from France and now based in Zurich, Switzerland, I possess a valid working permit that allows me to work freely across Europe.

Professional Experiences

• Data Scientist: UBS AG - Zurich, Switzerand

  Associate Director - Investment Bank Pricing Platform

  Since Sept 2023

  • Generating data using the pricing platform and developing a ML algorithm to predict the stochastic volatility charge from product economics and market data in order to decrease the pricing time of autocallable products and inrease the bank's competitiveness and market share.
  • Working on the integration of innovative products into the UBS environment to provide meaningful Data Analytics for Sales and Trading. Automated monitoring of the flow of priced and booked products allows not only to reduce the number of pricing errors by updating stale market data but also to drive business decisions and prioritize the team's book of work. The errors have already been reduced by 17% since January 2024 and multiple stakeholders acknowledge the progress.
  • I gather insights and statistics on past trade data to increase the amount of automated trades. I work on onboarding different regions and products on the fully automated flow.

  Graduate Program: Pricing and Product Modelling, IB Cross-Asset Structuring Platform, Credit Suisse AG

  Aug 2022 - Sept 2023

  • Designed and developed a scalable Python framework for live pricing and market intelligence content generation for sales and clients.
    Generated daily pricing screeners for sales desks across Europe and the US.
  • Designed and implemented a recommendation algorithm generating investment ideas on equity derivative products, incorporating the bank’s market outlook and statistical market data on more than 5,000 instruments.
    The recommender is available to the sales desks and professional clients on the global structuring platform.
  • Derived and leveraged competitors’ funding appetite from bond secondary market pricing screeners to dynamically adapt pricing parameters and globally distribute daily funding appetite graphs and time-series within the bank (distribution list including more than 200 colleagues).

  Graduate Program: WM Platform Product Distribution, Credit Suisse AG

  Jan - Aug 2022

  • Led the migration of a sales analytics application for FX, EQ, and FI sales desks to a new REACT-based tool.
    Defined tasks, priorities, and gathered continuous user feedback while proactively engaging stakeholders.
  • Managed the creation of a new platform feature for monitoring market events on user-selected financial instruments.
  • Gathered sales and trading data to support business decisions for the APAC platform.

• Data Scientist: Credit Suisse AG - Lausanne, Switzerland

  Intern - Global Information Barrier Solutions

  Sept 2020 - Feb 2021

  • Participated in the development of a monitoring platform for trading activities and potential suspicious behavior detection.
  • Created a Spark field dependency tool that identifies external dataset fields utilized in the entire transformation process to generate the final output.
    The algorithm also identifies output fields susceptible to changes, allowing monitoring of updates to external input fields.

• Data Scientist: Geosatis SA - Lausanne, Switzerland

  Intern - Analytics Team

  Summer 2017

  • Developed an algorithm identifying, using GPS data, frequently visited locations and commonly used paths for individuals wearing ankle bracelets to detect changes of behavior.
  • Dynamically rendered the results on maps that automatically adapt on behavior changes.

Skills

Languages

Education

• MSc in Data Science: EPFL - Lausanne, Switzerand

  2018 - 2021

  Main courses:

  • Machine Learning
  • Optimization for Machine Learning
  • Applied Data Analysis
  • Statistics
  • Artificial Neural Networks
  • Deep Learning
  • Data Visualization
  • Intelligent Agents
  • Project Management

• BSc in Communication Systems: EPFL - Lausanne, Switzerand

  2014 - 2018

  Main courses:

  • Algorithms & Data Structures
  • Algebra
  • Probability & Statistics
  • Calculus
  • Stochastic Models
  • Python
  • SQL
  • Java
  • Scala

Academic Projects

• Do Pedestrians Pay Attention? Eye Contact Detection in the Wild

  [Submitted paper]

  Spring 2021 (Master Thesis)

  Visual Intelligence for Transportation (VITA) - EPFL, Lausanne, Switzerland

  • Developed and published a state-of-the-art eye contact detection dataset for self-driving vehicles.
  • Used advanced deep learning and computer vision techniques to predict whether pedestrians are looking or not at the vehicle.
  • Generated annotation using AWS.

• Machine Learning based weighting strategies for climate models

  Spring 2020 (Master Semester Project)

  Swiss Data Science Center (SDSC) - EPFL, Lausanne, Switzerland

  • Used t-Stochastic Neighbour Embedding (t-SNE) to embed 40 climate models’ temperature time-series and compare it with Principal Component Analysis (PCA) and Multidimensional Scaling (MDS) for analysis.
  • Developed a new weighting scheme to address the limitation of assigning equal weight to climate models, thereby improving the accuracy of future temperature forecasts.

• Resilience of food sufficiency to future climate and societal changes

  [Preprint]

  2019 - 2020

  The Stanford Natural Capital Project - EPFL & Stanford, Lausanne & San Francisco

  • Employed decision trees to forecast the aggregated crops’ caloric yields at a granular pixel level.
  • The dataset is made of 1 million geospatial points capturing extensive information on soil conditions, temperature fluctuations, humidity levels, and a wide array of other climatic variables.
  • Evaluated future food sufficiency and resilience of countries and regions under diverse climatic and socio-economic scenarios.

• Multi-Application distributed ML on mobiles

  Fall 2017 (Bachelor Semester Project)

  Distributed Computing Lab (DCL) - EPFL, Lausanne, Switzerland

  • Distributed heavy computation tasks of Machine Learning across a universe of applications to a pool of mobile devices with varying memory capabilities.
  • Optimized the final training state of each model using an innovative approach.
  • Reduced the aggregated distance to the optimal training state of each model by a factor of 10.
  • Conducted a sensitivity analysis on mobile memory capabilities, ensuring stability and quality of the results.

Hobbies & Interests

🏉 When it comes to hobbies and things I love, I have a real passion for sports!
Being a huge sports fan, I can't help but get excited cheering on my favorite teams and catching up on the latest games.
I feel a special connection to Rugby, and I'm incredibly proud to have been a part of the Swiss 2023 champion team with Zurich.

⛷️ During the winter months, you'll often find me hitting the slopes, surrounded by the breathtaking mountain views.

🍳 And when I'm not embarking on data-driven journeys, one of my favorite things to do is cook up delicious meals. The joy of creating great food and sharing those moments together is truly something special to me.