Apache Spark applications in Fintech with Databricks and Wirecard

What we'll do

6:30 pm Food & Mingle Time
7:00 pm Introduction
7:05 pm Spark Payment Insights by David McKennell & Thomas
Mambrini / Wirecard
7:40 pm Break
7:55 pm Distributed Portfolio Risk Management using Spark by Nima
Nooshi / Databricks
8:30 pm Cake & Networking
9:00 pm Official End

Talk 1: Spark Payment Insights by David McKennell & Thomas Mambrini / Wirecard

The rudiments of advanced analytics in financial applications often include fraud detection. This presentation will give an overview on the means of tackling this problem using Apache Spark Streaming in conjunction with Kafka and a NoSQL database solution. The output of such analysis could be provided to alerting dashboards to create a near real-time mechanism for detecting anomalous transactional data.

David McKennell:
David is a data engineer at Wirecard, currently using Apache big data technologies such as Spark, Solr and Kafka to enhance their payment analytics platform. His specialisation is in Scala, which was developed during his tenure at Deloitte Digital in London, where he helped develop and maintain a worldwide Scala Play! application for a client in the UK public sector. David studied Actuarial Science and continues to use those skills in data science applications.

Dr Thomas Mambrini:
Dr Mambrini also works in the data products division of Wirecard, currently focusing on consumer analytics using Apache technologies. He previously worked at Banque de France where he developed machine learning algorithms to classify ‘risky’ changes within the IT infrastructure. He completed his PhD in the area of photovoltaic energy in Paris.

Talk 2: Distributed Portfolio Risk Management using Spark by Nima Nooshi / Databricks

Monte carlo simulations is the most commonly used methodology for estimation of the portfolio risk measures such as value at risk (VaR). To Backtest these models, it is usually required to fit the model on many rolling sets of time series data which can be pretty cumbersome. In this presentation, Nima Nooshi is going to talk about how the combination of bayesian generative models and spark distributed computing increases model accuracy and efficiency. Especifically, he discusses the advantages of generative models to estimate non-stationary, non-gaussian financial data distributions, and how different native distributed features of spark, such as pandas UDF and parallel hyperparameter optimization speed up the whole process.

Nima Nooshi is customer success engineer for strategic accounts at Databricks. He provides customers technical expertise for data science projects on spark. Prior to that, Nima was a customer facing data scientist at Accenture, where he managed and delivered data science projects to clients in the financial industry. He has also been actively involved in quantitative risk management practice when he was working for PwC. He also has a research background in physics and quantitative finance.

Please register for free by providing your full name (no nicknames allowed) and make sure you have your ID with you on the day of the event.
To ensure a smooth and fast check-in of all guests, access will only be granted if you have registered with your full name!

Registration for Wirecard Employees:
Access will only be granted if you have registered via Wirenet. No additional registration via meetup.com is needed.

*Disclaimer: By attending this event, you give your consent for Wirecard AG to publish any photos or videos that were taken of you at the venue on social networks.