Uber and Lyft release machine learning tools


It’s hard to a growing company these days that doesn’t take advantage of machine learning to streamline its business and make sense of the data it amasses. Ridesharing companies, which gather massive amounts of data, have enthusiastically embraced the promise of machine learning. Two of the biggest players in the ridesharing sector have made some of their machine learning code open source.

Uber recently released the source code for its Manifold tool for debugging machine learning models. According to Uber software engineer Lezhi Li, Manifold will “benefit the machine learning (ML) community by providing interpretability and debuggability for ML workflows.” If you’re interested, you can browse Manifold’s source code on GitHub.

Lyft has also upped its open source stakes by releasing Flyte. Flyte, whose source code is available on GitHub, manages machine learning pipelines and “is an essential backbone to (Lyft’s) operations.” Lyft has been using it to train AI models and process data “across pricing, logistics, mapping, and autonomous projects.”

Source: opensource.com