Dimas Lima Email & Phone Number

Rio De Janeiro, Rj, Br

Rio De Janeiro, Rio De Janeiro, Brazil

Leading a team of data scientists and data analysts focused on optimizing investment in promotions and vouchers to maximize growth with the best ROI and cost.To do this, we use causal inference, deep learning, and discrete optimization to model user behavior and subject all predictions to the constraints desired by business teams.

Rio De Janeiro, Rio De Janeiro, Brazil

Leader of a data science team focused on growth. The team's main goal is to discover how to optimize strategies using AI for communication and a better understanding of user behavior focusing on personalization and investment optimization to improve product engagement. main projects: Voucher sensibility based on causal inference, bringing over 20% cost per order reduction for the core business.Deep Learning Recommendation modeling approach answers three core questions: when should we send a notification, how many notifications should we send, and what communication should be used, based on the user's behavior and past preferences on app. The approach reduces the number of notifications sent at the same time it increases user engagement.

Rio De Janeiro, Rio De Janeiro, Brazil

Part of the team of instructors of the postgraduate course in data science at UFRJ. Main area of interest: recommender systems.

Rio De Janeiro, Brasil

Data Science Monitor in the Machine Learning Course of the data science specialization at UFRJ.

Rio De Janeiro, Rio De Janeiro, Brazil

Leading the data science squad at Globoplay, Sales Journey and Video Platform.As a leader, I am responsible for bringing technical insight, connecting business challenges with machine learning solutions, and assisting those being led in their individual development.Main projects: credential sharing and chargeback fraud modeling, next best offer for a set of globo products based on users behavior, churn modeling and CDN optimization focus on user's quality of experience.

Rio De Janeiro, Rio De Janeiro, Brasil

Optimizing content cache in Globo's Video Platform, focus on user's quality of experience and best use of infrastructure resources. It is done through AI and Machine Learning methods modeling both content metadata and time dependence and correlation in a high dimension scenario where algorithm's performance is a must to have. The machine learning team is responsible for building and deploying all models in APIs. Some models must be used in near real-time pipelines and others with daily update.

Rio De Janeiro, Rio De Janeiro

Worked in the growth hacking team at ge.globo. The team structure is defined by personalization, experimentation, advanced analytics, business analytics, and applied machine learning. My main responsibilities are to prepare and analyze big data to help businesses and introduce AI into the process of decision making. Some of the works that I’ve done are prospect prediction, user segmentation, high dimension model construction (when dealing with neural networks embeddings), and text and image analysis.

Rio De Janeiro, Rio De Janeiro, Brazil

Teaching Machine Learning and Data Science to new students.

Rio De Janeiro, Rio De Janeiro

At LabMA (Applied Mathematics Laboratory), I worked modeling mortality tables, conducting data analysis and data visualization. Also, as a research assistant, I was in charge to read and discuss academic papers with the students every week.

Rio De Janeiro, Rio De Janeiro

I worked as a research assistant under the supervision of two college professors in a health research that modeled psychological diseases to explain false positive diagnostics. We used Bayesian Inference applied to generalized linear model using Markov Chain Monte Carlo (MCMC) to estimate model parameters.

Master Of Science (M.Sc.), Statistics, A

My thesis focused on recommendation system applied to movies recommendation using the collaborative filtering approach and bayesian.

Bachelor'S Degree, Statistics

My final paper was focused on bayesian inference applied to generalized linear models using Metropolis-Hastings and Expectation.