My resume

Andresa Andrade



University of Campinas
BSc. Statistics 2010

Ryerson Continuons Education Program
Big Data & Predictive Analytics 2015


Programming languages: Python, SQL, R, Mongo, Linux, HTML
Spoken languages: Portuguese, English, Spanish, French

Logistic Regression
Decision Tree (Random Forest, XGBoost)
Linear Regression
Naive Bayes
A/B test


Paybright: Data Science and Analytics Manager

Manage a team of three members to deliver on data science and analytics projects. Some of the main projects that I led the team to deliver or was responsible to deliver are:

  • Deliver an enhanced fraud model using XGboost methodology in Python.
  • Design our credit engine as a Microservice in AWS, translating our SQL stored procedures into Python and separate our external API calls in a separate service.
  • Deliver on customer satisfaction analysis using NLP on our customer survey to understand main opportunities of improvement.
  • Work with the business (specially our client support and resolutions team) to automate daily desks and improve unit costs.
  • Leverage data to investigate customer fallout rates.

Paybright – Data Scientist

 (January 2019 – December 2019)

  • Design our Fraud engine as a microservice in AWS parallelized with our web application system. Work closely with the data engineer and Devops Engineer to automate our data science model deployment. 
  • Developed and implemented our fraud detection model using logistic regression using machine learning and data wrangling libraries such as pandas, sklearn, numpy, statsmodels etc.
  • Develop our data pipeline for analytics infrastructure using SQL.
  • Automate our email marketing referral program using Python.
  • Support the business in automation for reports.

PC Financial and Services – Data Scientist

(May 2018 to December 2018)

My role at PC Financial involved working closely with our marketing, product and digital team to find opportunities for improvement in our processes. Some of the projects I have worked on are:

  • Develop a new way to calculate efficiency of marketing channels using a multi-touch channel attribution focused on customers instead of focused on interaction. Most of the work was done using a combination of the following languages: SAS, python and SQL and using the following machine learning methodologies and libraries: Gradient Boosting, sklearn and pandas.
  • Using a holistic view of marketing opportunities and the analysis with the multi touch attribution model, created a formula to estimate our media mix and maximize the business ROI.
  • Work with the digital team to measure effectiveness of our web experience. Owned our testing framework to develop new features and measure success. For small design tests, a simple A/B test framework would work but for larger and complex experiences, we have more complex models, such as a cluster model to understand customer interactions with our platform.

TELUS – Data Scientist

(February 2017 to April 2018)

As a Data Scientist at Telus I used statistics and machine learning algorithms to support the marketing team. Some of the projects I have worked on were:

  • A customer segmentation model based and assigning a revenue potential for each customer. The model used K-means to group our customers. This helped our sales team to improve their business strategy and to increase our customer satisfaction from 65% to 78%.
  • A product penetration analysis to support both our product and marketing teams with channel planning and price forecast.
  • A sentimental analysis to read our customer survey feedback and categorize into negative and positive. The model also had a drill down in the subject mentioned in the answer, i.e. if a customer’s feedback is related to price, the model used the text mining to include it in the price category.

TELUS – Senior Marketing Analyst

(November 2015 to January 2017)

My role at Telus related to our digital initiatives and to support the teams with measurements and strategies. Using tools such as our Hadoop, Domo, Adobe Workbench and Adobe Analytics. I supported the marketing team by connecting datasets and by creating meaningful insights for their daily decision-making processes. Some of my accomplishments were:

  • The creation of a measurement and reporting process for our Display and Paid Search activities. These reports summarized the main KPI’s relevant to each campaign as well as the insights that they brought to the business.
  • Our offline attribution project: this project consisted in connecting our offline sales to our digital media activity. Google and Facebook were key partners. This enabled the team to have visibility in the end-to-end effect of marketing in the customer journey.
  • To improve our cost per acquisition (CPA) in 150% for wireless. It was part of my role to assess our digital investments and understand the optimization opportunities. Some of my recommendations involved but were not limited to audiences, tactics, offers, publishers and time of day.
  • To implement an online attribution report. I worked with Adobe to implement in Adobe Workbench a report that contained a 90 days attribution for all the online traffic sources that a customer would use before a purchase. This not only supported our team to efficiently invest our digital budget but also predicted the channels that are more relevant for our customers based on different offers.
  • To develop a model that correlates offline sales and online behaviour. We learned that the first step of our checkout flow has 70% correlation to online sales.

Sears Canada – Web Analyst

(August 2013 to August 2015)

At Sears my main challenge was to build a data structure that would drive the decisions made by the business. Some of the problems that I faced and had to overcome in order to achieve my goal:

  • Work and integrate datasets from multiple sources. At Sears, I had the opportunity to work heavily with R, Python, MySQL, SAS and Hive. These technologies were used not only to manipulate, transform, predict data but also to automate some of the manual work done among my team. I also worked with some well used models such as Multilinear Regressions, Time Series, K-means and A/B tests.
  • Producing outputs that could be interpreted by the business owners. I had the opportunity to create some fundamental tools for our teams, one good example was a matrix that showed the items that had to be online in order to sell in store. This way our buyer’s team could use it to prioritize their work.
  • Create a correlation model that understands the relationship between different touch points in and final conversion. One of our findings relates the number of visits before someone buys a specific product. Having this metric enabled the email marketing team to fire specific offers in specific days with a customized message.
  • Create an attribution model that assigns different weights for Online Marketing Channels based on user behaviour.
  • Build segments and associate revenue opportunities to them. For example, we found some segments in our database that were more focused on buying furniture and these people would be more likely to buy in store.
  • Improve our SEO program using a predictive model that would use variables such as seasonality, type of keywords, product and a few other variables to predict increments in Sales.
  • Present key discoveries to Sears Executives.

Rocket Internet GmbH – Business Intelligence Analyst

(January 2012 to September 2012)

I was hired to drive the business analytics for our 12 start ups. I ended up being focused mainly on two of them (Zocprint and 21Diamonds). Below are some of the tasks I was responsible for:

  • Maintain the Google Analytics API and help to develop the Facebook Ads API.
  • Create visualizations for our online reports and share them with our investors.
  • Run A/B tests on our online campaigns.
  • Support the CRM program with my super skills as a Statistician.
  • Keep track of Customer behaviour and understand any abnormality.
  • Implement data layers for our web application tools.