Worked in an AI research group (Group with Strategic Focus on Intelligent Systems at ITESM) tackling the combinatorial optimization NP-hard problem VRP with RL and DL.o Currently in the process of publishing approach “Combining Constructive and Perturbative Deep Learning Algorithms for CVRP”.o Created open-source high-performant implementation for the paper “A Deep RL Algorithm Using Dynamic Attention Model for VRPs” using Pytorch and Numpy.o Increased model speed by +400% over the previous public implementation and reduced its memory complexity for training by a factor of the number of nodes in the VRP using a memory-efficient gradient trick.

Worked on Facebook-Instagram achieving the following goals:o Implemented computer vision algorithm on ad delivery flow for enhancing low-resolution product images to generate more value for Instagram.o Developed and designed pipeline and service usage for the enhancement flow of products where optimizing queries/second was a critical component.o Leveraged multiple code bases, APIs and programming languages (c++, hacklang & sql) to create services in pipe.

Worked as an Engineering Practicum Intern on the User Location Reporting team on the area of GEO at Google acheiving the following goals:o Worked in the ULR team on the reliability of Location History support for Incognito in Google Maps.o Created on-device end to end test for the team that automates and simulates human dependent and device processes to ensure reliability on ULR and Incognito mode.o Developed new memory tests for ULR that run whenever a new feature is created to flag significant deviations in memory usage from Google Play Services with such feature.o Collaborated in Incognito storage migration of Google Maps along Google engineers.