Juan Diego Alfonso Ocampo

As a Data scientist at this internship I collaborated closely with product owners, software engineers, and data scientists within an Agile SCRUM team to developed a SaaS prototype to automatically interpret and standardize heterogeneous Bill of Materials. Applying Machine Learning (ML) and statistical methods to reduce BOM interpretation and processing transforming a multi-hour task into minutes. Key Results Reached ~80% accuracy in BOM interpretation applying Machine Learning (ML), NLP, embeddings, and statistical methods. Designed and deployed a modular cloud architecture including: - Rust (backend services) - Python (PyTorch, Scikit-learn) for ML models - PostgreSQL database - Docker for containerization and AWS for deployment Delivered a robust, and scalable prototype. Validated robustness across real-world heterogeneous datasets. Produced full documentation, integration tests, and knowledge transfer for maintainability.

As a Data Engineer, I worked in a team of five within the R&D department. We collaborated closely with researchers and scientists from the Stability, Analytical Chemistry, and Product Development teams. My primary role was to develop and deploy cloud-native data pipelines to ingest, transform, and visualize biochemical testing data stored in Oracle databases. I created automated ETL workflows using SQL, Power Query, and SharePoint-integrated pipelines to support daily scheduled refreshes via background services. A key focus was to standardize data structures, reduce reporting latency, and enable seamless cross-team access to high-integrity data for analysis. I also led the design of modular BI architectures supporting versioned deployment of dashboards used in both production and experimental workflows. Key outcomes - Built interactive Power BI dashboards for specific analyses and nutrient degradation tracking, reducing manual data handling by 67%. - Designed reusable data models and parameterized fact tables, improving dev team efficiency and report scalability across shelf-life datasets. - Streamlined quality control workflows by 40%, replacing outdated Excel reports with dynamic dashboards reflecting 15+ years of historical data. - Ensured high data integrity and optimal query performance by optimizing SQL logic, aligning schemas, and applying clustering for scalable filtering. This work enabled the R&D teams to make data-driven decisions on product formulation, shelf-life validation, and regulatory compliance (EFSA) in real time.

As a Data Engineer at Globant, I played a key role in building and optimizing data architectures for various client projects, in the transformation of data to feed predictive analysis algorithms or the creation and transition to data-based architectures. Main Responsibilities: Design, build, integrate, and maintain data processing infrastructure. Work closely with data analysts and data scientists to assist in the transformation and preparation of data for analysis. Implement ETL tools using PySpark to move data between different storage systems. Optimize database queries and conduct code reviews to ensure the efficiency and quality of work. Evaluate and propose new tools and technologies to improve the efficiency and scale of data processing. Work with Cloud tools such as GCP and AWS to integrate data-based solutions. Key Achievements: I led the restructuring of a successful migration of our data warehouse to a cloud solution, resulting in a 20% reduction in operational costs and a 30% increase in query speed with the optimization of parametrizable jobs. I designed and developed 15 robust and scalable data pipelines that feed a Data warehouse and machine learning applications, improving the accuracy and efficiency of the models by 10%. I collaborated in the creation of an internal framework for CI/CD process improvements, adding ETL tests, reducing production errors by 25%. I delivered internal training on best practices in data engineering, raising the quality standard and technical knowledge of the team.

As a Data engineer we worked on the creation of an infrastructure to store the big data retrieved from the company improving the 50% the operational cost of the old infrastructure based on a Data lake and fed by automated worksheet files, we created an infrastructure based on the cloud, orchestrating the ETLs process with Airflow, Kubernetes to process the Data manipulation with Spark and Kafka to streaming data of some transactional databases.

Participation in the management and solution of first level incidents for software distribution by Microsoft for the home.

Management and solution of user incidents, first-rate analysis regarding the problems with the platforms developed by the company, verification and analysis of the database for the accompaniment of doubts regarding customer information.

We participate in the administration of UNIX servers and storage management for data centers, as well as incident management for 4 clients, implementing Solaris, RedHat and Debian, as well as administration and implementation of storage and backup tools such as DataProtector and NetBackup.