Verified Expert in Engineering
Data Scientist and Developer
With a strong mathematical background (a master's degree in mathematics), Nicolas is a passionate data scientist who can contribute the ideal combination of machine learning knowledge, practical programming skills, and a problem solving and analytical mindset to a project. He has a demonstrated history of transforming business problems into data-driven solutions and recently has worked as a data scientist at the global insurance company, Allianz.
Jupyter Notebook, RStudio, Git, Linux
The most amazing...
...thing I've coded was an R package to predict life insurance claims based on individual characteristics. It is a novel approach going beyond the status quo.
Data Science Lead
- Led the development of data science and data engineering projects within the domain of clinical trials.
- Set up a cloud infrastructure for model training, deployment, and application prototyping which increased the impact and visibility of our team within the organization.
- Designed and administrated a Neo4j graph database to centralize organizational datasets and leverage graph algorithms to answer complex business questions.
- Developed an interface to the graph database allowing non-technical users to ask questions in natural language. We trained a deep learning model to translate English to Cypher (graph query language).
- Created an optimization algorithm to select the best possible sites for clinical trials and oversaw the roll-out and integration into the business operations.
- Supported the external adjudication process of two medical studies.
- Automated the process of combining, filling, and sending a large number of PDF forms using Python.
- Kept track of the data exchange with the adjudicators via an automated Excel table and provided a dashboard to create progress reports.
- Used Selenium to automate downloading and gathering PDF files from a website, which would have been weeks of manual work.
Focus Sensors Limited
- Examined the implementation and the mathematical concepts of an extensive codebase of an anomaly detection algorithm of sensor data.
- Changed the core architecture from static data files to streaming data using Kafka.
- Tested and optimized the new architecture in terms of processing time and output integrity.
- Implemented and productionized a personalized machine learning algorithm to classify transaction data using the AWS SageMaker and Lambda infrastructure.
- Detected trends in the customers' behavior and created frequent reports to present the results, which have been published regularly on the company's website.
- Completed various data analyses and POCs to answer requests from the business using a combination of SQL and Python for the back end and Jupyter Notebooks and Plotly to present findings.
- Analyzed the microloan data to identify the relevant features that have an impact on repayment behavior.
- Implemented and tested a Python module that returns a credit risk score together with a detailed explanation.
- Deployed that module on the AWS SageMaker and Lambda infrastructure to fully integrate it with the current system.
Sopra Steria España
- Developed new methods to measure business success for a retail client and its implementation in Python.
- Performed a post-analysis of retail promotions using SQL and Python and presented findings to stakeholders.
- Optimized SQL queries to extract insights from large tables.
- Restructured and optimized an internal Python package to extract and visualize statistics of large database tables.
Data Scientist (Master Thesis Student)
- Wrote my thesis about machine learning methods to model life tables.
- Performed the preprocessing, analysis, and modeling of data with a size of over 100GB.
- Built and tested an R package for internal usage in the actuarial department.
- Conducted my final presentation in front of experts as part of the official training series.
- Implemented exhaustive performance optimization of R code using vectorization, parallelization, and optimized packages.
- Implemented and supported extensive interactive data-driven dashboards in R-Shiny.
- Developed a product recommendation system for corporate clients based on the clients' characteristics and product history.
- Built a productive automated system for the early detection of problems with products or business processes based on client complaint data.
- Implemented the visualization of complex data and presentation of insights using Plotly, D3.js, and R Markdown.
- Performed topic modeling and text mining of client complaint texts using LDA.
- Constructed presentations concerning theory and programming packages within the field of machine learning.
- Created internal programming packages to streamline and simplify frequently used data science tasks.
Fraunhofer Institute for Industrial Mathematics ITWM
- Worked on the project Senrisk (Senrisk.eu/), which predicted price fluctuations of corporate and sovereign bonds based on news sentiments.
- Built recurrent neural networks in PyTorch to predict financial time series.
- Developed statistical methods for fraud detection in the health insurance industry.
- Implemented a Python package for financial time series prediction, including the integration to a web service.
- Constructed a software prototype in R Shiny to visualize the impact of different sample sizes in the context of fraud detection.
Universidad Técnica Federico Santa María
- Implemented software in C# to evaluate financial options based on the Black-Scholes model.
- Created a detailed report about the theoretical foundations of option price valuation.
- Conducted research related to the Black-Scholes model and financial time series.
EU Project SENRISK
My part was mainly the implementation of the prediction system. We used recurrent neural networks and boosting methods and built a Python package to streamline the whole process.
Analysis and Visualization of WhatApp Chatshttps://github.com/l47y/whatsappalytics
Machine Learning Demonstration Toolhttps://github.com/l47y/ml_tool
Android App Course Analyzerhttps://github.com/l47y/SiCourses
On the main page, you have an overview of all courses, and you can export and import a list of the courses. Finally, you can see statistics of the evaluations, and a map shows the places where the courses have taken place with some additional information about it. Currently, it is only available in Spanish.
Python, R, SQL, C#, Markdown, Kotlin, HTML, CSS
RStudio Shiny, LightGBM, Selenium
Pandas, Ggplot2, XGBoost, Keras, Scikit-learn, Beautiful Soup, NumPy, PySpark, PyTorch, SciPy
Plotly, Amazon SageMaker, Amazon Athena, Git, LaTeX, Sublime Text, Spyder, Microsoft Excel, Amazon QuickSight, PyCharm, Microsoft PowerPoint
Data Science, Business Intelligence (BI), Automation, Testing
RStudio, Amazon Web Services (AWS), Linux, Azure, Jupyter Notebook, Anaconda, Databricks, Apache Kafka, Docker, Amazon EC2, Dataiku
Data Analysis, Dashboards, Data Analytics, Applied Mathematics, Mathematics, Dashboard Development, Data Visualization, Machine Learning, Random Forests, Dashboard Design, Reporting, Data Reporting, Financial Markets, Financial Data, Big Data, Algorithms, Process Automation, Optimization, Web Scraping, Web Crawlers, Time Series Analysis, Statistical Analysis, Neural Networks, Credit Risk, Loans & Lending, Software Engineering, Data Engineering, Android Development, Natural Language Processing (NLP), Streaming Data, Signal Processing, Large Language Model (LLM), Project Leadership, Microsoft Office, GPT, Generative Pre-trained Transformers (GPT)
Redshift, Databases, SQL Server Management Studio, MySQL, MongoDB, Neo4j, Database Management
Master of Science Degree in Financial and Actuarial Mathematics
Technical University Kaiserslautern - Kaiserslautern, Germany
Spent an Exchange Year in Financial Mathematics
Universidad Técnica Federico Santa María - Valparaíso, Chile
Bachelor of Science Degree in Mathematics
Technical University Kaiserslautern - Kaiserslautern, Germany
Big Data Fundamentals with PySpark
Applying SQL to Real-world Problems