David Jaramillo

Científico de datos con conocimiento en Python, R, Spark, Hadoop, AWS y mucho más. Con gran interés en aprender nuevas cosas y hacer parte de un entorno lleno de retos y nuevas experiencias.

Científico de datos (SURA Asset management):

• Trained, tuned, and deployed multiple time series model using LSTM neural networks, Prophet, and statistical techniques to predict the value of the company’s funds in the different countries, using the autoregressive component of worldwide market indexes as well as the previous value in time of the fund.

 • Performed statistical simulation to come up with the optimal ways to construct the different analytical tools for the investing area of the company. Doing things such as finding the best way to optimize the calculation of prediction intervals for the LSTM model. 

• Worked and developed multiple ETL processes to extract data from different sources coming from 6 different countries in the continent, transform it into a readable and usable format and finally load it into a BigQuery database in order to make it available for other areas and projects such as modeling and reporting.

 • Segmentate different risk factors regarding money laundry prevention using advanced clustering techniques to properly understand and describe the general profile of clients, products, states, etc. according to several variables of interest using python and machine learning libraries. 

• Create multiple python programs to detect unusual transactions from clients, based on the clusters in which they were classified and their respective financial information. • Analyze datasets related to the behavior of clients, products, and jurisdictions to generate programs that can effectively define the risk level of a specific individual, based on different variables.

Analista Cuantitativo (Peo Partners): 

• Backtested and participated in the definition of up to 20 different investing and asset allocation strategies using risk factor exposure to come up with several portfolios formed by 200 selected stocks that can outperform public equities like the S&P500, using R and Python. 

• Studied and analyzed the behavior of factors like value and profit of more than 500 of the most renown stocks in the US and European markets to find patterns and relations that may allow the company to more optimally define advanced investing strategies to construct new portfolios, using R and Python. 

• Shared presentations with insights about the resultsfound in the backtesting and analysis of the different selected stocks that form the basis of the 20 seen strategies, through the use of financial metrics and data visualization using R and Python.

Profesional en Estadística de la universidad nacional de colombia.

Certificado en:

• Deploy Machine Learning models on GCP + AWS Lambda
• Big Data with Apache Spark and Python
• Python for data science and machine learning bootcamp