Matthias Darblade, Algorithms Developer in Buenos Aires, Argentina
Matthias Darblade

Algorithms Developer in Buenos Aires, Argentina

Member since November 12, 2018
Matthias is an actuary with over six years of experience in machine learning. He was the chief data scientist in a multinational company—leading AI projects in eight countries. The types of projects that Matthias are looking for would ideally involve deep learning, analytics, and data-related tasks.
Matthias is now available for hire


  • Chewse
    Algorithms, TensorFlow, Pandas, Scikit-learn, SQL, Python 3...
  • Prosegur
    TensorFlow, Pandas, SQL, Python 3, Artificial Intelligence (AI)...
  • 2x3
    Amazon Web Services (AWS), Algorithms, Pandas, SQL, Python 3, AWS, Python



Buenos Aires, Argentina



Preferred Environment

Pandas, Jupyter, Python

The most amazing...

...application I've built was the one where I implemented reinforcement learning for cost optimization.


  • Lead Data Scientist

    2018 - PRESENT
    • Worked remotely as the acting lead data scientist for a Series C startup based in San Francisco.
    • Created the core business optimization model for supply matching. The model uses a similar architecture to Google's AlphaGo and was written in Python with C++ binding.
    • Maintained high-code quality through code reviews, automated tests, and continuous integration.
    • Composed several reports and insights to improve supply matching using graph theory, statistic inference, and machine learning.
    Technologies: Algorithms, TensorFlow, Pandas, Scikit-learn, SQL, Python 3, Artificial Intelligence (AI), Machine Learning, Python
  • Corporate Head of Data Science

    2017 - 2019
    • Led the churn-reduction program with an objective of a 20% reduction in churn across eight countries.
    • Oversaw the development of the machine-learning algorithms and management of external resources to design and implement the final architecture.
    • Created a machine-learning algorithm to improve the mobile application of the company. This algorithm understood client behavior to remind them of actions they might have forgotten to do.
    • Supervised the hiring, building, and leading of a team of three data scientists. Led a team of five consultants based in Spain.
    • Developed a financial analysis to justify capital investment into data-science projects.
    Technologies: TensorFlow, Pandas, SQL, Python 3, Artificial Intelligence (AI), Machine Learning, Python
  • Consultant

    2018 - 2018
    • Created a machine learning algorithm for pricing for a Chilean startup.
    • Deployed the ML model in AWS.
    Technologies: Amazon Web Services (AWS), Algorithms, Pandas, SQL, Python 3, AWS, Python
  • Freelance Developer

    2018 - 2018
    • Created a recommender system for an Indian YouTube competitor.
    • Used NLP and a neural network to give a list of recommendations for videos to look for.
    • Built a model that worked in six languages, including Hindi, Bengali, Tamil, and Telugu.
    • Deployed the model in AWS for online prediction.
    Technologies: Amazon Web Services (AWS), Algorithms, Pandas, SQL, Python 3, Artificial Intelligence (AI), Machine Learning, AWS, Python
  • Senior Data Scientist

    2017 - 2017
    Telefonica de Argentina
    • Worked on a machine learning model to define where new antennas should be deployed. The input of the model were 200TB of data representing the monthly traffic of more than 20 million clients.
    • Developed ETL processes in Spark and Hadoop, along with the development of the ML model (training and testing).
    • Deployed the model in production using SparkML and Docker.
    Technologies: Algorithms, Pandas, SQL, Python 3, Big Data, Machine Learning, Hadoop, PySpark, Spark
  • Global Data Scientist

    2014 - 2017
    BNP Paribas Cardif
    • Defined and developed a dynamic pricing library for automobile insurance in Chile.
    • Performed R&D at the data laboratory of the head office in Paris (NLP, deep learning, and so son).
    • Combined artificial intelligence and behavioral economics to automate claim payments.
    • Built a tool to improve quarterly closing. The time for closing went from one month per quarter to four days per quarter.
    • Improved a reserve calculation algorithm to not depend on human interactions for predictions.
    • Automated the back-testing of several finance algorithms for the quick development of solutions.
    Technologies: Pandas, SQL, Python 3, Machine Learning, SAS, Python
  • Lecturer

    2015 - 2015
    Universidad de Buenos Aires
    • Pitched and lectured a course about machine learning for actuary students in one of Latina America's most prestigious universities.
    • Taught various concepts of data science and Python to students.
    Technologies: Pandas, Python 3, Machine Learning, Python
  • Actuary

    2013 - 2014
    • Consulted with various clients on actuarial science and portfolio analyses.
    • Segmented a health insurance company portfolio to predict the financial impact of a new regulation and gave recommendations to clients as to what type of product to develop.
    • Led the yearly update for a product of Addactis PM Export and coordinated and tested the development of the software with the engineering team.
    • Gave talks about the use of machine learning to learn about client behavior.
    Technologies: Pandas, SAS, Microsoft Excel, Python


  • Development of a Real-time Pricing Strategy

    I developed a web scraper for an online insurance business in Chile and created a model to react to changes in competitors' prices for each segment.

    The model combined game theory, behavioral economics, and machine learning to bring profitability to the car insurance industry, which is known for having an extremely low return on investment.

  • Supply Matching Algorithm

    I built out a supply matching algorithm for a startup using Python and C++ bindings. Using Monte Carlo Tree Search combined with a fast and slow neural network evaluator, the system determined the best possible move to improve overall COGS and reduce errors.

  • Recomender System

    I created a recommender system for an Indian online video sharing platform. The system worked making recommendations at the end of each video with an API response time of 45 ms and was able to recommend based on context instead of pure similarity of content.


  • Languages

    Python, Python 3, SAS, SQL, Go, C++, Excel VBA
  • Frameworks

    Spark, Hadoop, Flask, Django
  • Libraries/APIs

    PySpark, TensorFlow, Pandas, PyTorch, Keras, Scikit-learn
  • Tools

    Microsoft Excel, Periscope Data, Jupyter
  • Paradigms

    Data Science
  • Industry Expertise

    Insurance, Healthcare, Retail & Wholesale
  • Other

    Artificial Intelligence (AI), Big Data, Natural Language Processing (NLP), Risk Models, Machine Learning, Graph Theory, Schedule Optimization, Algorithms, Cosmos, Cryptocurrency, AWS, Neural Networks, Deep Learning
  • Platforms

    Linux, Amazon Web Services (AWS)


  • Master's Degree in Actuarial Science, Finance, and Risk Engineering
    2012 - 2014
    ISFA | Institute of Financial Science and Insurance - Lyon, France
  • Bachelor's Degree in Mathematics and Management
    2009 - 2012
    Université Claude Bernard Lyon 1 - Lyon, France

To view more profiles

Join Toptal
Share it with others