Machine Learning Engineer at Tres Astronautas

RESPONSIBILITIES

● Design, develop, and deploy an end-to-end intelligent document system using Python to extract valuable information from shipping labels.

● Implement computer vision models using Python-based libraries and frameworks to build AI-powered parcel and pallet dimensioners for clients with 5k+ users.

● Optimize models for production performance using tools such as ONNX and TensorRT.

KEY ACCOMPLISHMENTS

● Designed and implemented a highly scalable Python-based machine learning pipeline that handles 2M+ shipping labels per day, by optimizing the inference speed of state-of-the-art multimodal models.

● Increased freight flow by 100 scans/day by deploying a high-performance on-device computer vision model, helping to increase the operational volume by 500k+ packages per day.

● Mentored the team to follow best practices in developing efficient and maintainable AI-powered solutions, resulting in a decrease in the cycle time from 2 months to 1 month.