Developed an advertising frame that achieves a high engagement rate and click-through rate, utilizing the provided images and text that will comprise the advertising frame. Accomplished by leveraging computer vision techniques and an AI multi-agent framework called Autogen.  

Retrieval Augmented Generation (RAG) system with different strategies  including a simple RAG approach, a multi-query approach, and using AutoGen. Evaluation using RAGAS.

Build Contractor Advisory Bot using React for frontend and Flask API in the backend.

An automatic prompt generating system. By leveraging a Retrieval-Augmented Generation (RAG) pipeline, the system is able to generate high-quality prompts based on the user's query.

Improves context retrieval by generating alternative prompts on top of the user's original question. And evaluated the generated prompts.

Used the CausalNex library to uncover the structural relationships in environmental factors like weather and holidays, as well as the day of the week (weekday vs. weekend) and the distance between the customer and driver locations.

Also used DoWhy library to estimate causal effects, quantify causal influence, and perform root cause analysis. 

Conducted backtesting and applied modern portfolio theory, 

Used Moirai-1.0-R-large and Chronos models to handle time series forecasting tasks.