AI Research
The field of Generative AI is still relatively new and has many challenging problems to be solved. Banavo.AI is focused on solving the difficult problems of combining predictive and generative AI based on fine-tuning open source models with both structured and unstructured data. Banavo.AI is dedicated to making LLMs readily available to both developers and enterprises, so they can unleash their true potential.
Banavo.AI has a close collaboration with the University of Texas, Austin through eminent professors and their students in the space of Generative AI, Neural Networks, and Predictive Modeling from the Department of Computer Science and Department of Statistics and Data Science.
Our AI Research Areas
Research focused on Customer Adoption of Gen AI Technologies
01
Training and Fine-tuning LLM Models to play specific roles and perform assigned tasks
This research is focused on integrated and automated learning frameworks for LLM Models for performing eCommerce tasks. Integrated learning frameworks will seek to combine RLHF techniques and LLM Model fine-tuning using LoRA to improve the accuracy of LLM Models and track their performance.
02
Embedding Co-Pilots in eCommerce Operations
Marketing and Sales Business Process Automation using declarative business rules is the current state of Art. Gen AI LLM Models and Predicitiva Analytics is about to change how tasks are accomplished in Gen Work Flow which get orchestrated in real time. Our research focusses on Goal based composition, execution and optimization of business processes using LLM's.
03
Deploying and Operating LLM Based Co-Pilots in Secure Production
DevOps to MLOps to LLMOps are all converging to accelerate the embedding of ML and LLM Models into software applications and workflows. Currently there are no standards on how to do it. Banavo.AI is pioneering a set of technologies to bring ML and LLM Ops together with DevOps and Infra Ops to simplify deploying Gen AI Apps into Production using a Declarative Language Framework.
04
Economics of Deploying LLM-trained Co-Pilots in the Enterprise
Co-Pilots enabled by LLMs and ML models are powerful capabilities to boost productivity in the enterprise. However, various studies show that this productivity benefits are not without significant cost. Banavo.AI is driving research around LLM-driven Activity-Based Costing on the business side and combining it with cost of operating Co-Pilots in the AI provider clouds or on premises. The goal is to provide CTO, CIO and CEO the total cost of ownership and ROI model of Gen AI Technologies.