Portfolio
A curated set of AI applications, machine learning systems, MLOps platforms, and data products based on real production experience and selected research work.
Frontend / Creative Portfolio
Built a modern photography portfolio experience to showcase visual work with fast navigation, responsive layouts, and a polished project presentation layer.
Impact: Demonstrates product-minded frontend engineering through a real, public-facing portfolio project with strong visual presentation.
MLOps / ML Platform
Designed and implemented an ML automation platform to standardize training, deployment, and lifecycle operations for machine learning workflows.
Impact: Production-oriented platform work for repeatable ML delivery.
LLM / Recommendation Systems
Built recommendation systems using large language models and vector databases to improve cross-selling opportunities and product discovery.
Impact: Applied LLMs to real business recommendation workflows.
MLOps / Production ML
Implemented shadow and A/B deployments for machine learning models, improving validation, monitoring, and production rollout safety.
Impact: Safer model releases through monitoring and controlled experimentation.
Computer Vision / Edge AI
Designed and deployed computer vision systems on Jetson devices for vehicle movement detection, route analysis, and curb detection.
Impact: Distributed edge inference for real-world video and geolocation workflows.
NLP / Research
Researched deep learning architectures to predict which of two press headlines would become more viral, comparing LSTM, BERT, and transformer encoder approaches.
Impact: Published research applying deep learning to headline virality prediction.
Machine Learning / Web Application
Built a data-centric application to predict student dropout risk and cluster dropout patterns using supervised and unsupervised learning techniques.
Impact: Award-winning software engineering grade project at Universidad Icesi.
LLM / Data Automation
Fine-tuned GPT-3 models to automate data labeling workflows, reducing manual effort and improving the scalability of ML data preparation.
Impact: Practical LLM automation for internal ML data workflows.
Data Generation / AutoML
Created synthetic datasets to train and evaluate Google AutoML NLP and Vision models, improving model accuracy through targeted data generation.
Impact: Improved model training data for NLP and computer vision use cases.