Next Generation Machine Learning Platform
Increasing collaboration, efficiency & productivity
Originate is a concept-to-launch digital acceleration partner headquartered in San Francisco with additional offices in Los Angeles, Newport Beach, New York, Las Vegas, and Berlin. Originate works with clients to accelerate innovation, partnering with well-funded startups and forward-thinking industry incumbents turn to when they need to go from zero to one rapidly and reliably. Originate combines world-class engineering and design to deliver high-quality products that create real value, real fast.
- Founded in 2007 - Brent Cohen, CEO, Jeff Scheinrock, CFO
- Midsize Company
- 150+ clients
Originate’s machine learning engineers, Danny Meyer and Amanda Doucette, work in tandem on research-based machine learning projects. Their work requires data exploration, model development, and a lot of collaboration. Prior to Spell, they were running machine learning experiments on one shared machine, which slowed their ability to iterate quickly.
They used Slack messages to share status updates and would need to find one another in the office to review and discuss model performance. It wasn’t efficient and the overhead was reducing productivity. A colleague at Originate told them about Spell, an end-to-end machine learning platform that handles infrastructure and environment setup and management. They signed up and were immediately impressed. With Spell, they were finally able to focus on real machine learning work instead of infrastructure set-up. A machine learning DevOps solution was exactly what they needed.
Moving to Spell changed the way the Originate team worked. They no longer needed to spend time setting up and managing their infrastructure. They were able to start experimenting faster and seamlessly move from idea to training a model. And with Spell’s advanced machine management tools, they could switch cloud providers and access a wide variety of machine types.
Spell’s collaborative platform allowed them to monitor and view projects without needing to find one another in the office. The platform created visibility and transparency on the team. Resources are shared with ease and the team is able to be more creative since they’re able to more easily build off each other’s ideas and experiments. “We love the ability to see what’s going on,” said Danny.
"If you have a client project, I don’t know a faster way to get up and running than Spell,"
In just a few months, the team has trained hundreds of models on Spell and is using the platform for every client project.
When the team needed to switch cloud providers from AWS to GCP, Spell had the tools that made the switch fast and easy. Machine management is effortless. Spell automatically spins down machines once they’re no longer needed. And easy access to Spot Instances saves money. The team uses Spot Instances for experimentation and then switches to on-demand machines when they need them. Spell’s auto-backup feature also means they don’t have to worry about losing work. With all these improvements, their compute costs are down 65%.
Spell has not only increased the team’s efficiency, productivity, and revenue but has eliminated the operational hassle and limitations the Originate team was experiencing before. Being able to collaborate and view and monitor each other’s experiments has made working together easier. They can quickly try new approaches, like distributed training, and with Spell’s metric and logs dashboard, they can stop experiments that aren’t performing well early on.
Machine Learning Projects with Spell
Request a Demo
Schedule an in-depth demonstration with a Spell representative to learn how Spell can help streamline and accelerate your machine learning development.