Happy new year! One year ago we were a team of 6 working to launch the first version of our product to a small number of beta users. Now, after a year spent building with and learning from the customers in our closed beta, growing our team to 15 people (and counting!), and making lots of product additions and improvements, we’re gearing up for a broader launch. More on that soon 🤫. In the meantime, keep reading to learn what we’ve been up to in the last month.
Build with BaseTen 🛠
Here are a few ways you can start using BaseTen today:
- Restore an old, damaged, or low-quality photo with GFP-GAN (+ See more impressive examples of restored photos on the @basetenco Twitter)
- Test and iterate on your model servers with scaffolds
- Kickstart your ML project with a pre-trained model from our model zoo
- Build an audio transcription app with wav2vec
While we’re still in closed beta, we’re eager to get more of you using BaseTen for free. If you don’t have access to BaseTen yet, you can jump to the top of our waitlist by telling us a little about how you want to use BaseTen here.
See what’s new on BaseTen 📣
We share new features and product improvements on our changelog. Here are a few things we’ve shipped in the last month:
- Install system packages: Specify system-level packages that you want to install in your Python environment.
- Optimized base images: We’re now periodically updating Docker base images for every model framework we support. We’re seeing that model deployments can be 2-5 times faster as a result.
- Delete test data: Start with a clean slate by deleting all the rows from your BaseTen-provided Postgres tables.
We also released a handful of navigation changes, all intended to make it easier for you to get around the product and differentiate between exploration mode (when you need to understand how all your models, apps, and data relate to each other) and focus mode (when you’re actively building in BaseTen):
- Main navigation: The main navigation minimizes when you’re editing worklets and views, to give you more space to build.
- Applications and models pages: We replaced cards with tables, which are better suited for larger numbers of models and applications and set us up to support filtering and sorting in the future.
- Data page: See your BaseTen-provided Postgres tables and external connections in one place.
Join our team
We’re hiring! We posted a number of new roles including forward-deployed engineer, developer advocate, and technical writer. Visit our careers page to see all our open roles. And if you don’t see a role that fits but are excited about what we’re building at BaseTen, we’d still love to hear from you.
You can also read more about what it’s like to join BaseTen—our very own Samiksha wrote about her experience in Part 1 - Working at an early-stage company as an early-stage engineer.
Choosing the right horizontal scaling setup for high-traffic models
Horizontal scaling via replicas with load balancing is an important technique for handling high traffic to an ML model. Let’s examine three tips for understanding how to properly replicate your instances to save users time without wasting your money.