José Conceição

My own Youtube channel where I make content about programming :-)

Working for the Barcelona Data Science Hub in the financial division of Novartis. Software Lead of one of the most mature products of the entire hub. - Successfully migrated product from on premise to cloud. - Complete refactor of the entire project to meet production quality standards(~15000 lines of code). - Maintained and updated CI/CD pipelines. - Documented the entire project on confluence, software description, run book and several How-to articles. - Workshops on several Technical Tops like cloud, and core programming concepts. - Provide support to Data Scientist with SE best practices.

Projects: - Acessed how useful Quality Estimation for Machine Translation is, published paper with https://eamt2020.inesc-id.pt/proceedings-eamt2020.pdf (in conjunction with my coworkers), led the training of several QE models(Openkiwi framework) and their respective deployment via Rest API (flask + docker) and experimental data collection. - Several experiments related to Machine Translation, such as grid search,Teacher-Student models, incremental training and more. (Marian, OpenNMT,Linux) - Development of several Tag Handling solutions for MT. - Trained SOTA production ready MT models . (continuous) (Marian, OpenNMT) - Lead developer of a multilingual chatbot project (English,Spanish,Chinese,Arabic). (Rasa, Docker, Docker-compose, Flask, Diagflow) - Redaction project (fast-api + spacy) - Deliver of several internal tools via small web applications using flask. - Creation and Administration of DynamoDB + ElasticSearchService - Automation of model training, procedures and workflows - Constantly trying out new technologies and see where they could be used in the business Using AWS daily for launching instances testing new services and creating new products using the Amazon Stack

Ubiqum Code Academy is a an innovative educational company that simulates a real istic technology projects in web developpment and data science coworking space (WeWork). Instead of the traditional academic/lecture methodology , they practice a project-based learn-by-doing methodolgy where we conduct a sequence of analytical activities culminating a presentation to a simulated group of stakeholders. MSc of 800h ("learning by doing" methodology) dealing with: -Client segmentation -Cross Selling -Recomendation systems -Profitability of new potential products -Forecasting - Applied Data Science to Wifi indoor localizaiton project . -Sentiment Analysis with AWS Tech tools Rapid Miner , R , python , SQL