We are a research lab that specializes in developing custom deep learning architectures and algorithms with a focus on pushing the boundaries of efficient inference.

We solve real-world problems with novel applications of existing techniques, whilst also exploring new techniques and new applications of existing techniques of machine learning. We are currently focused on the following areas:

• Fast and efficient training and inference of diffusion transformers for real-time video synthesis
• Lightweight and small text-to-speech (TTS) models for low-latency and low-bandwidth applications
• Applications of generative transformer encoders for graph-based long-long-term-memory (LLTM)
• Bitnet-like low-bit high-performance large language models
• Custom small language models for specific tasks and domains

What can you do with AI?

Have a problem that you think could be solved with AI? We are always looking for new challenges to tackle and new problems to solve. If you have a problem that you think could be solved with AI, we would love to hear from you to discuss research opportunities and funding.