Elton Vieira

Data Scientist @Cisco

San Jose, CA, US

Research and development of Machine Learning models with the goal to improve the business in different ways. Experience with the entire ML-focused development pipeline:• requirements collection• data (pre)processing• delivery of updates to stakeholders• deployment of new models via API endpoints (using Python, Django and the Google Cloud Platform)Some of the projects I\'ve worked so far:1) Knowledge Base Search Engine: • Creation from scratch of a search engine utilizing industry-standard NLP concepts (word and sentence embeddings, topic modeling, etc.) • Development of a learning-to-rank model that re-ranks the search results by relevance, utilizing as training data the users’ feedback (both implicit and explicit). This model is based on gradient boosting techniques (Python library used: XGBoost). • Best answer retrieval and search snippet generation utilizing BERT transformer models (hosted on the cloud using the GCP AI Platform).2) Task Distribution Optimization: • Development of an in-house algorithm that optimizes the distribution of technical tickets to our group of pre-sales engineers. As new tickets are created by our customers, this system is capable of routing them to the best engineer available (in terms of expertise, availability, etc.), thus improving important business metrics such as time to assignment, number of customer-initiated escalations and customer satisfaction. • This system utilizes a collection of both static rules and ML to distribute the workload among the team in an intelligent way. Techniques utilized: Ensemble of classification models based on Naives Bayes and Gradient Boosting (utilizing the sklearn package), and CP-SAT Solver (utilizing the Google OR Tools package).

Universidade Tecnológica Federal do Paraná

Telecomunication Systems