• Implementation of a transformer-based forecasting algorithm for predicting transactional volume time series at multiple levels of temporal granularity. This solution generates forecasts up to two weeks in advance for intervals of 15 minutes, 1 day, and 1 week. In each case, achieving results with an R² metric above 0.9. • Ideated, designed, and implemented a correlation-based model (leveraging mutual information among commercial transaction time series) to identify quantifiable relationships for “Athia,” a multimodal generative LLM. This solution provided data-driven “reasoning seeds” for AI agents to explain variable behaviors, using multi-threading and GPU-based processing. It processed data from 15+ variables for KFC Ecuador in around 10 seconds.

• Developed an algorithm (Deepstream) to analyze security cameras in self-checkout areas for retailers such as Walmart Mexico and OXXO. The algorithm generates alerts for possible theft situations, such as people paying for only a portion of their products or passing by without paying. It was capable of processing up to 10 cameras in real time per Nvidia Jetson Xavier, with each camera operating at around 5 FPS. Additionally, the alert delay time on the platform was less than 10 seconds.

• Optimized the product prices for TBC Corporation across both physical and digital outlets by leveraging data analysis, estimating cross elasticities, and developing a decision algorithm for their price strategy

• Developed an autonomous algorithm (SQL & Python) for e-commerce products prices recommendation. Switching from a manual system with a latency of up to 6 months to daily price changes. • Implemented Machine Learning (Python) models to evaluate the annual change in the product assortment, increasing the range of products considered by 200%. • Optimized a code (SQL & Python) for competitors' promotions price recognition, making it 12 times faster and increasing its accuracy from 50% to 80%.

• Accomplished a consultancy for the "Digital strategy for chemical products" of Virutex-Ilko company, creating a roadmap of the indispensable projects for its digital transformation towards the 4.0 industry. • Developed an AI algorithm (Python) in IoT devices to replace the fault detection system of a Torre S.A. production line. Switching from faulty manual registration with 10-minute delays to an automated method of vision algorithms connected to a unified database with real-time data availability with a maximum delay of 1 second and 95% accuracy.