Come work on Artificial Intelligence & Parashift Platform

We develop and implement Intelligent Document Processing in the technologies used by numerous of people every day. We believe that an approach based on advanced AI for Computer Vision and Natural Language Processing, supported by increasingly more efficient AI training hardware, is the only way to achieve a general solution for truly autonomous document processing at a global scale.


Tweek the algorithms

Develop new algorithms that read documents by creating a high-fidelity representation of documents. To train neural networks to predict such representations, create algorithmically accurate and comprehensive ground truth data by obfuscating information from real-world documents and enabling data augmentation and synthetic data generation. Use state-of-the-art techniques from representation learning, active learning, transfer learning, and regression constrained learning to develop a robust system that works on complex and messy real-world documents. Evaluate your algorithms at the level of the entire customer base and leverage swarm intelligence.

Neural networks innovation

Drive research and apply state-of-the-art peer research to train deep neural networks for problems related to the classification and extraction of documents. Our networks analyze raw documents and images to perform semantic segmentation, table detection, and various extraction mechanisms. In doing so, the networks learn from billions of data points, a wide variety of business contexts, and the most diverse document variations in the world, continuously coming in from different industries and companies.


Codebase improvements

Throughput, processing time, accuracy, and determinism are the main criteria by which we optimize our codebase. Build Parashift Platform software foundations from near the lowest levels of the tech stack to tight integration with our custom software landscape and optimize the KPIs.

Platform evaluation infrastructure

Build comprehensive evaluation tools and infrastructures to compare numerous models, perform meta learning, and accelerate the pace of innovation, track performance improvements, and prevent regressions. Write code and generate real-world-like data that feeds the Parashift Platform for live debugging or automated testing.


Hardware improvements

To do anything here, unfortunately, we don’t have the budget (yet). However, the goal would be to eventually be able to make chips that power and enable our Intelligent Document Processing software from the ground up. All of this is to take full advantage of all of the architectural and microarchitectural enhancement possibilities and to get the maximum performance out of the hardware. Once again… Not something we can put you on today. No matter how much we would like to. But we are working on it. ????

Come help us build the future of Intelligent Document Processing

By continuing to use this site you consent to the use of cookies in accordance with our cookie policy.