Profile Details

Contact

★★★★★

☆☆☆☆☆

USD 35 /hr

★★★★★

☆☆☆☆☆

Hire Alison D.

Brazil

USD 35 /hr

Data Scientist / Machine Learning Engineer at Globo.com

Profile Summary

Since undergraduation, I've specialized in the area of Artificial Intelligence by studying specialized disciplines and developing models of facial recognition and segmentation of bathers, always trying to create models that use the maximum of all the best possible technologies. I wish to be able to work in this area that I decided to specialize within the university and to test my skills in the job market in real situations. Nowadays I'm in the big data experimentation team and I have the responsibility to organize and do the whole process of AB testing and rollout at globo.com. Making a platform for experimentation, as well as creating and implementing performance metrics. I am still responsible for building reinforcement learning algorithms to optimize certain areas at globo.com like ads viewability. I still already developed computational models for facial recognition, creating a face verification system for condominium entry and crowd in a football stadium. My Open Source Projects: https://github.com/alison-carrera Projects - Research (Private) Search Algorithms for Multi-Agent Patrolling Problem Jul 2015 – Jul 2016 Project descriptionThe Multi - agent Patrolling (MAP) is the problem to coordinate a group of "n" agents for a given environment as a graph or a network during a period of time "t". The agents move through the edges to visit all the nodes of the graph repeatedly, so as to minimize the maximum interval between visits. This project addresses the use of search algorithms to solve the MAP, since there is a gap in the literature about this approach. As a result of this work, a library of generic search, able to solve any problems, since they are adapted to it it was developed. But in particular, this library was used to solve problem instances of great form Patrol as planned. These optimal solutions have been integrated into a simulator of the patrol, allowing comparisons with other algorithms (usually heuristics) to the problem. They were also made comparative analysis of the optimal outputs with some search algorithms and built a GUI for yaps simulator, in order to improve the viewing solution and to facilitate their understanding. Crowds Segmentation for Bathers Monitoring System Focusing on Low Computational Cost Feb 2016 – Dec 2016 Since 1992, sixty people were attacked by sharks on the beaches of the metropolitan area of ​​Recife and all twenty-four died. There are many signs of warnings on the beaches of the region, but most swimmers ignore the risk. The State Committee Sharks with Incident Monitoring and the Secretariat of Social Defense as a means to avoid more possible accidents of this kind, have sought solutions that they can reduce the incidents of sharks in the region. Treatment of images with bathers (crowds) is a very complex task due to the great ambiguity of the scenarios, the presence of unimportant objects, a varying background and visual occlusions by the own swimmers, which may be with body parts immersed in water. Having this information as this project aims to develop a focus with crowds of image segmentation algorithm in low computational cost using boss regions detection algorithms.

Subject Matter Expertise

• ☆ Facial Recognition Technology
• ☆ Data Analysis
• ☆ Data Processing
• ☆ Big Data Analytics
• ☆ Machine Learning
• ☆ Comparative Analysis
• ☆ Python

Services

• Data & AI

Work Experience

Data Scientist

globo.com

August 2018 - Present

Data Scientist

Pegasus Technology

March 2016 - July 2018

Education

Bachelor in Computer Science

Federal Rural University of Pernambuco

December 2012 - January 2016

Certifications

• Certification details not provided.