Christian Lopez

I built a vector database for products and categories using MarqoAI with multilingual LLM models, improving similarity searches and customer experience on the site. Using neural collaborative filtering, I developed the product recommendation system for the company, the implementation of which increased the retention rate and allowed the launch of cross-selling campaigns for specific groups of customers according to their purchasing behavior. I built the key database model for the development of microservices apps and visualization metrics, using a structured and relational PostgreSQL model. This allowed for greater security, maintainability, traceability and, fundamentally, much more efficient data clarity and consumption, greatly reducing infrastructure costs. I built complete visualization panels for each provider. With statistical analysis and fundamental metrics of their sales, retention and conversion rates, inventory, etc., allowing their performance analysis and strategy improvement.

I designed an AI agent, using the RAG methodology, capable of designing guides for the creation of digital content and ad campaigns, based on vectorized search on pieces of content on social networks with high engagement. I integrated AI tools such as YOLO, AssemblyAI and HumeAI to preprocess social media content, segmenting it into images, audio and speech. Through sentiment analysis and image processing, I extracted the key characteristics of content with a high level of engagement. I developed ETL processes and ML models with the complete life cycle in the AWS ecosystem with business data: customer and product management, demand prediction, risk analysis, marketing strategies, social media content, ad campaigns, profitability, prices, investment, etc. I developed complete software solutions (backend and frontend), integrating internal and external APIs for the automation of product management, products and orders, customer service, ad campaigns and social media content, drastically reducing operational costs.

I developed software for real-time monitoring and forecasting of sensors, with live visualization, anomaly alerts, and breakage prediction, improving operational stability and production line efficiency by reducing downtime by approximately 15%.