We're looking for breadth-first engineers ready to tackle a variety of challenges across our stack—from backend scaling to frontend performance to programming language theory. You'll play a critical role in evolving the nascent Serenade product into an essential tool for developers around the world.

Here's a sample of projects you'd be able to work on from day 1:

• Build on top of open-source technologies like Tree Sitter and the Language Server Protocol to iterate on our generalized code transformation platform.
• Implement new voice commands that seamlessly automate even the most complex workflows.
• Meet with developers using Serenade in their daily workflows to understand what works well and what doesn't, then build solutions.
• Create integrations with developers' favorite tools and iterate on our cross-platform Electron app.
• Scale our containerized backend to handle a rapidly-growing number of developers using Serenade.

And, here's a bit more about our product stack:

• Our backend consists of modern, containerized, JVM-based, gRPC services on top of AWS technologies like RDS, ElastiCache, and S3. We also use Rust for high-performance code parsing.
• Our frontend is an Electron-based app written using TypeScript and React alongside native node.js addons and cross-platform components written in Rust (e.g., for Win32 and X11).
• Serenade integrates with a variety of apps via their respective plugin architectures, using languages from Python to JavaScript.
• Whenever possible, we contribute back to the open-source community.

We're looking for machine learning engineers to iterate on our cutting edge ML systems, from our speech-to-code engine to our code prediction and transformation platform. You'll play a critical role in evolving the nascent Serenade product into an essential tool for developers around the world.

Here's a sample of projects you'd be able to work on from day 1:

• Architect transformer-based neural networks to model both code and natural English.
• Apply state of the art speech recognition techniques to our dataset to optimize the accuracy and performance of the product.
• Build on our training pipeline, which includes automatic labeling and seamless backtesting.
• Use semantic parsing techniques to enable deep understanding of user intent.
• Support a wide range of speakers and environments (e.g., noisy backgrounds, speaker identification, and so on).

And, here's a bit more about our ML stack:

• Our speech-to-code engine uses state of the art deep learning served using a containerized, gRPC streaming service written in C++ for maximum performance. We've also developed our own models trained specifically for programming.
• We use AWS services to automate our training pipeline, along with TensorFlow and PyTorch.
• Whenever possible, we contribute back to the open-source community.