Lucas Pavanelli Email & Phone Number

Rio De Janeiro, RJ, BR

Los Gatos, California, United States (Remote)

• Led the development of a product called FineTune, allowing users to select their dataset and fine-tune for different NLP tasks (MT, ASR, text classification) using suppliers such as HuggingFace, OpenAI, and Azure.
• Worked on ACL 60-60 initiative by translating and evaluating over 10K ACL papers into 60 languages.
• Developed Automode, a tool that automatically selects the best MT and ASR suppliers. Improved performance and reduced financial cost when compared to six suppliers.
• Developed ASR classification models to detect age and Arabic dialects.
• Co-authored three scientific publications together with the Science team.

Los Gatos, California, United States (Remote)

Rio De Janeiro, Rio De Janeiro, Brasil

• Worked on a resume parsing tool applying machine learning algorithms (e.g. HMM, gradient boosting) to segment a CV into sections like Personal Data, Work Experience, Education. (Java and Python)

Rio De Janeiro, Rio De Janeiro, Brazil

• Course: Advanced Data Structures.
• Taught binary search tree, red-black tree, hash table, graph algorithms and other related subjects. (C)

Rio De Janeiro, Brazil

• Jobzi is a tech start-up based in Rio de Janeiro that aims to provide smarter ways of matching jobs and candidates using Machine Learning techniques.
• Implemented search algorithm using Solr. Improved search time in 50% and results quality in 30%.(Java and Python)
• Used NLP techniques to recommend skills for resumes. (Python)
• Improved company’s infrastructure using microservices and applications such as ELK, Sleuth and Zipkin. Decreased build time in 100% and debug time in 50%. (Java)
• Worked on job’s salary identification using gradient boosting and metrics such as precision and recall to evaluate proposed method. Improved recognition from 70% to 90%. (Java and Python)

Rio De Janeiro E Região, Brasil

• Developed a stochastic simulation system for Petrobras, under the orientation of Eduardo Laber and Hélio Lopes. Allowed them to save hundreds of thousands of dollars in the investigation of wells to be drilled. (C/C++)
• Participated in a machine learning project aimed at improving the selection of attributes to split data in decision tree. (Python)

Rio De Janeiro Area, Brazil

• Course: Introduction to Engineering
• Taught basic C language and how to build a Windows Form application.

Rio De Janeiro Area, Brazil

• Developed documentation and back-end tests of a TOEFL training application called EduSynch. (Ruby on Rails)

Rio De Janeiro Area, Brazil

• Developed publicity web pages. (HTML, CSS and Javascript)

Master Of Computer Science, Information Extraction, Gpa 9.9/10.0

Dissertation: An End-to-End Model for Joint Entity and Relation Extraction in Portuguese

Bachelor Of Engineering (B.E.), Computer Engineering, Gpa 8.2/10.0

Undergraduate thesis: Segmentation of unstructured textual data using HMM. Developed an algorithm inspired in the Viterbi algorithm to.

High School With Technical Degree, Programming