Piotr Dariusz Antoniak - CV

Piotr Dariusz Antoniak - CV




Master’s Degree; Pompeu Fabra University and Universitat Autònoma de Barcelona, 2020-21.

Thesis: Predictive power in online investing chatter. (Keywords: BERT, Big Data, forecasting, neural networks, NLP, sentiment analysis, scraping, stock market, social media, WallStreetBets)

Bachelor’s Degree; Aberystwyth University, 2017-20.

Thesis: Evaluation of the post-crisis monetary policy in the USA using Taylor rules. (Keywords: monetary policy, Taylor Rule, Zero-Bound)


Macroprudential Policy at ECB; 2022 - ongoing :

  • Development of a top-down reverse stress testing model that allows to simulate thousands of scenarios and test multiple parallel constrains (solvency, leverage, liquidity) of Eurozone banks. It utilizes exposures to interest rates, sovereign bonds, economy on industrial levels and unconventional monetary policy. Model’s results are included in bi-yearly ECB’s MacroPrudential Report and presented in front of all EuroArea Central Banks Financial Stability/MacroPrudential representatives;
  • Fulfilled ad-hoc data requests from various, highly granular databases using SQL;
  • Management of data pipelines utilized by members of my teams.

Financial Services Business Model Development Praktikum at BMW (references), 2021 – 2022:

  • Statistical Analysis, Modeling, Predicting, and building databases tailored for use cases and providing basic GUI / executables to present in front of stakeholders.
  • Deploying Minimum Value Project ML models to AWS, using Lambda Services, EC2 and SageMaker.

Project I worked on:

Fraud detection for car dealerships:

  • explainable AI for predictions delivered by classification Machine Learning model,
  • visualizing network for personal information (such as email or address),
  • deploying minimum value project into Amazon Web Services using Lambda Functions and EC2,
  • building basic Graphical User Interface and executables then presenting in front of stakeholders.

Building a search and retrieval system for internal data drive and confluence page:

  • Deep Learning pipeline based on Sentence Transformers, Maximum Inner Product Search and Question Answering model.

Customer feedback:

  • mapping out most popular topics using Sentence Transformers and Zero-Shot classification,
  • worked on how to avoid people being unhappy about certain issues,
  • implementing a series of Natural Language Processing models for feedback classification.

Customer Lifetime Value:

  • Developing Machine learning model to predict how much the customer will spend on the next car purchase (a mix of regression and classification task)

2017 – 2020:

  • Part time job in hospitality – barista, bartender, kitchen porter. One year in charge of the evening kitchen.


Using Google.

ML/DL/Cloud computing:

  • Languages: Python, MATLAB, SQL
  • Libraries: pandas, numpy, matplotlib/seaborn, pytorch/tensorflow, scikit-learn, scipy, pymc
  • Tools: Git, Amazon Web Services, MS Office, Bloomberg API
  • Statistics/Econometrics/ML/AI: Generalized Linear Models, Random Forests, Boosting, SVM, LSTM, RNN, ConvNets, Transformers, MLP, Explainable-AI, Causal Inference, Time Series

Side projects

Predictive power in online investing chatter.

  • My Master Thesis.
  • I investigated millions of comments on investing/trading forums in order to extract sentiment and check whether users retail traders are so unskilled as claimed by various studies.

Open Domain Question Answering – repo. Working demo.

  • A tool that can answer questions using documents given to it.

Simple chatbot from scratch.

  • A simple chat bot that uses predefined question-answer pairs;
  • It’s simplicity is based on the fact that extending usage is just a matter of updating a .csv file.

GPT-2 twitter bot mimicing CentralBanks

  • A Twitter Bot based on GPT-2 further pre-trained on speeches from major Central Banks.


  • Stanford/DeepLearning.Ai: 1,2,3
  • MITx CS: 1,2
  • Other: 1

Additional Information

  • Triathlete,
  • T Maelgwyn Davies Prize - Best Economics Dissertation (Aberystwyth, June 2020),
  • Experience trading alternative assets (2017-2018) and options (2019-2020),
  • Awarded the Second Place in The Export Challenge organized by the Transport Department (Swansea, November 2018),
  • Mentor at DeepLearning.ai