Research Scientist

We are revolutionizing how users interact with large language models (LLMs), similar to how search engines transformed the early internet. As the number of AI models grows, it’s becoming increasingly challenging to determine which model is best suited for specific tasks. We address this by offering a model router that selects the optimal model for a given prompt in real-time, ensuring high performance and cost-efficiency.

Our approach involves predicting model performance without execution, allowing us to consistently match users with the most effective solution. The goal is simple: remove the complexity of AI model selection so you can focus on building great products.

This model-routing tool is just the beginning. By developing techniques to better understand model behavior, we aim to tackle the broader issue of ensuring that AI models behave in predictable and reliable ways.

About the Role:

As a Research Scientist, you will contribute to groundbreaking work that seeks to improve our understanding of AI models. A primary focus will be on refining and scaling a technique known as "model mapping," which translates transformer models into more interpretable representations, such as programs. You will develop innovative methods to make transformers more understandable and study their behavior in this newly mapped domain.

Responsibilities:

• Design and conduct experiments to assess model mapping effectiveness
• Investigate how AI models can be transformed into interpretable programs
• Manage and analyze large datasets from interpretability studies
• Explore the internal workings of large language models (LLMs)
• Develop new approaches to understand and interpret the behavior of LLMs

AI Research & Content Development:

• Produce technical research on model interpretability and AI model routing
• Write research papers for top AI conferences and journals
• Create blog posts to explain complex AI concepts to broader audiences

Technical Community Engagement:

• Participate in technical discussions with the AI research community
• Represent the company’s perspective in forums and at conferences
• Build relationships with key influencers in the AI field

Ideal Candidates Will:

• Be passionate about improving AI interpretability and advancing AI tools
• Have a strong desire to discover algorithms that drive intelligence and are committed to understanding how models work
• Thrive in a fast-paced startup environment
• Have experience with ML algorithm implementation (e.g., PyTorch) and distributed training (e.g., PyTorch Lightning, DeepSpeed)
• Be interested in mechanistic interpretability and write clean, well-structured code

What We Offer:

• Competitive salary and equity packages
• Comprehensive health, dental, and vision insurance
• Unlimited PTO
• Daily lunch and regular team dinners