Machine Learning and Modeling
You have lots of complicated data, but don't know how to optimize it for use in the real world. We have helped startups, small businesses, and multi-national corporations design Machine Learning and Computer Vision models to do just that.
Systematized model training and cross-validation for a neural net model applied to drone imagery on agricultural experiments. We then enabled this same process on a new crop by fine-tuning the same model on new labeled data.
Trained a novel version of Yolov4 to do bounding box detection for images taken at a different angles relative to open-source dataset imagery. Resulted in >90% mAP on the testing dataset.
Embedded System Design
You need complicated systems and algorithms that work in the real-world. We will help guide your decisions about the right hardware platforms and implement software optimizations that improve speed, memory consumption, and reliability.
Developed BIST (Built-In-Self-Test) for a satellite prototype to ensure system memory stability while deployed in space.
Converted observational study machine learning algorithms into a real-time system. One computational unit went from taking ~40 seconds to < 1 second.
Wasting time, money, and energy chasing inconsistent data around your organization? We build systems that turn the data all over your organization into robust business intelligence.
Designed a central database combining data from 5 different sources across the business syncronizing them all in one location.
Built an autonomous system to detect changes in the status of a customer's tax return and to send them timely, custom-tailored emails from the appropriate tax representative.
Created a custom tool to instantly extract all media files belonging to Griffin Road Media from a data server whose web interface only allowed for clients to individually retrieve files.