AI Research at Intuit

Controlling the generation of language models through prompt optimization, in-context learning, supervised fine tuning for improving personalization, accuracy, and reliability of model responses.

Enhancing the language model’s knowledge and reasoning capabilities through external knowledge injection, grounding, and neuro-symbolic approaches.

The fintech domain requires more than just the data used to train models. We also often need substantial subject matter expertise to make the best decisions. This knowledge might come in various forms, like rules or laws written in naturallanguage or storesd in multiple knowledge bases.

Improving the efficiency and quality of language models for practical applications.

Developing enterprise AI native applications that can harness the powers of generative AI, traditional AI and machine learning, and tools and APIs for solving customer problems with high accuracy and robustness.

Developing algorithms for information extraction from different types of documents, interleaved generation of text and images, design and layout optimization for text and images.

Developing conversational agents for end-to-end customer assistance for question answering, task completion, recommendation, and advice.