Ruben D.

Key Skills

Freelancer Overview

Throughout my four years of experience as a Data Scientist, my approach to AI training has evolved from fine-tuning base models to architecting end-to-end Machine Learning pipelines (MLOps). I have actively designed and implemented predictive and Natural Language Processing (NLP) models, optimizing hyperparameters and structuring custom loss functions within Python and TensorFlow environments using SQL/NoSQL architectures. My expertise extends beyond training classification and regression models; I ensure that the data flow—from ingestion via ETL processes to production inference—is both scalable and efficient. Regarding data processing and labeling, my role has centered on strategic curation and quality control within a Data-Centric AI framework. Moving beyond passive oversight, I have designed automated frameworks for bias detection, outlier imputation using advanced Exploratory Data Analysis (EDA), and cross-validation to mitigate overfitting. Recognizing that the performance of any neural network or supervised learning algorithm depends directly on the purity of its "ground truth," I have led the structuring of labeling taxonomies and the cleaning of large volumes of structured and unstructured data, transforming raw data into production-ready assets.