Kin Fai Kan

Title: Kin Fai Kan and Trust Layers for Large Language Models.

Introduction

Kin Fai Kan is an inventor based in Palo Alto, California. His patent work includes a cloud-platform approach for adding configurable safeguards to interactions with large language models.

Latest Patents

Kan is associated with 1 patent, titled “Trust Layer for Large Language Models.” The invention describes a cloud platform that receives a client’s prompt and applies the client’s configuration parameters to modify it before sending it to a large language model. After receiving the model’s response, the platform can provide it to another model that estimates the likelihood that the response contains content from specified categories. The platform can then transmit the response, those probabilities, or both to the client.

Career Highlights

The patent presents a way to place configurable controls between a client and a large language model. Its approach combines prompt modification with analysis of generated responses, giving clients information that can help them assess potentially sensitive or categorized content.

Collaborations

Kan’s listed coworkers include Shashank Harinath and Eugene Wayne Becker. Their names provide context for the professional relationships associated with his patent work.

Conclusion

Kin Fai Kan’s patent focuses on a configurable cloud-based trust layer that modifies prompts and evaluates language-model responses, connecting his work to safeguards for AI systems.