Álex Filipe Santos

My Projects

AVDExtractor
2018 at Stone Co.
NLP tool to extract top positive and negative keyphrases from an employee’s performance evaluation.
Technology: Python - FastText, gensim, NLTK, spaCy, NetworkX, pandas, NumPy
batistest
2018 at Stone Co.
Python module for mass processing and evaluating different ML classification algorithms and generating performance reports.
This tool was crucial for improving turnover classification models by ~10 pp using feature selection, cross-validation and hyperparameter optimization methods.
Technology: Python - scikit-learn, matplotlib, seaborn, pandas, NumPy; to be open-sourced soon.
Integration SDKs and APIs
2019 at Stone Co.
Python modules for bulk API integration (data fetching and parsing) with external sources relevant to People Operations.
Internal database schemas for employee data centralization from different sources and ETL interfaces.
Technology: Python - Flask, Apache Airflow, requests, asyncio, concurrent.futures; some SDKs to be open-sourced soon.

InspiraSonho
2015-2018
Student opportunities platform that empowered 12,000+ high school and undergraduate students across Brazil in their professional and educational development.
Technology: JavaScript + jQuery, PHP, MySQL, Apache
LabStocker
2014-2015
Cloud-based inventory management system for chemistry labs, including a dashboard and preliminary predictive statistical & machine learning based methods for storage consumption.
Technology: Java, JavaScript + jQuery, PHP, MySQL, Apache
Honors: Selected among the top 3 Chemistry projects countrywide at the Brazilian Chemistry Congress (2014).
Publications: Software patent (Brazil) issued in 2017. Patent number: BR 51 2016 000345-6.

Evaluation of Circadian Time Series Analysis Methods Using Simulated Data
2016 at Amherst College
Engineered simulations and algorithms to test efficiency of a number of circadian time series analysis methods using MATLAB.
Honors: Awarded as Outstanding Poster for Biomathematics at the Joint Mathematics Meetings 2018 in San Diego, CA. See JMM 2018 poster.
Efficiency of Modified Clays in Water Softening
2015 at Federal Institute of Rio Grande do Norte
Used statistical and numerical methods to evaluate performance of different chemically modified clays for water treatment.
Publications: Conference paper published in Congress Proceedings of the XVth IWRA World Water Congress in Edinburgh, Scotland.
Scorpion: Treatment System for Water Reuse
2015 at Federal Institute of Rio Grande do Norte
An adaptable, scalable and low-cost device for water treatment, based on interchangeable modules with filtration materials.
Experiments were performed with activated bentonite and vermiculite clays, demonstrating the materials as potential treatment agents.
Publications: Product patent (Brazil) issued in 2017. Patent number: BR 10 2015 010742-0 A2.