Open Source Platform
Compiled, automatically parallel python for data science
What is the platform?
Ufora's open source platform automates the engineering code required to run single-threaded programs in parallel across a cluster of machines. It JIT compiles your Python so it's fast on a single thread, automatically detects opportunities for parallelism in your programs, and then distributes those threads and the data they access across the available RAM and CPU's in your cluster.
What it's Not
The Ufora platform solves similar problems as paradigms like Hadoop, Spark, etc., but unlike those systems Ufora doesn't require you to work within any framework (e.g. Map Reduce) or language API's (e.g. PySpark). Our goal is that you can work natively in Python, using your existing workflow, and get significant performance benefits.
Open Source & Support
Ufora's platform is an open source project licensed under the Apache 2.0 License. We're eager to hear your thoughts about this early version of the product and what additional functionality you'd like us to prioritize. Learn more on our documentation page.
Infrastructure Consulting to make your models faster and more accurate
You're best equipped to analyze data about your own business. We have the expertise to design and engineer distributed systems infrastructure that allows your data science to scale. Let us handle the infrastructure so you can focus on the analysis.
Our engineers provide Data Science Engineering consulting to select firms to optimize their data science stack for speed, scale and accuracy. Building on years of experience in building highly complex distributed computing systems and parallel processing engines, Ufora's engineers can quickly identify opportunities for efficiency in your existing data science stack and implement the changes without disruption to your ongoing data science work.
Our engagements have typically yielded 5-100x speedups, and substantial accuracy improvements. We can also help you streamline your business processes to get the answers you want out of your data quickly and efficiently. We get you acquainted with best-of-breed technology alternatives, including open source projects, and we typically work in Python, R, and C++.
"Our model accuracy improved significantly after Ufora sped us up and enabled us to work with 10X more data." Hedge Fund CTO
Who is Ufora?
Ufora is a team of computer scientists who wanted to tackle the challenges of distributed computing. Braxton, Ronen, Tom and group of faithful friends first came up with this new approach a few years ago and began experimenting with ways to automatically parallelize simple scripting code. The critical design goal was finding a way to reason about scripting code to figure out opportunities for parallelism of threads and data on the fly. It took years of experimentation and many false starts but now we know it's possible and we’re excited to bring it to the world to help make it a reality.