Vinicius Bastos Gomes

Created image-based similarity models for approximately 20 million products - Developed text-based similarity models for around 4 million brands - Improved/created brand feature scores using techniques such as custom calculations and distributed machine learning models - Implemented tooling for reliability of machine learning models' lifecycle, including model versioning and experiment tracking Tools: Python, PySpark, EMR, Docker, Databricks, Elasticsearch/Open Search, Airflow, Faiss, MLFlow, Postgres

- Developed a Machine Learning algorithm for predicting barley fields with high probability of exceeding the permitted concentration of Deoxynivalenol. - Modelled and deployed a fully automated multi algorithmic (Machine Learning) commodities trading system. - Created several spark ETL pipelines. Main tools: Python, Spark, Kedro, Docker, Databricks, MLFLow (MLProject).