A Serverless Architecture for Real-Time Stock Analysis using Large Language Models: An Iterative Development and Debugging Case Study
This work provides a practical, incremental solution for individuals seeking to democratize AI-powered financial tools, focusing on iterative development and debugging in a serverless context.
The paper tackled the challenge of building a low-cost, real-time stock analysis system by designing a serverless architecture that uses the Gemini API for qualitative assessment, automated data processing via GitHub Actions, and a static frontend, resulting in a near-zero cost operational model with a publicly accessible application and open-source code.
The advent of powerful, accessible Large Language Models (LLMs) like Google's Gemini presents new opportunities for democratizing financial data analysis. This paper documents the design, implementation, and iterative debugging of a novel, serverless system for real-time stock analysis. The system leverages the Gemini API for qualitative assessment, automates data ingestion and processing via GitHub Actions, and presents the findings through a decoupled, static frontend. We detail the architectural evolution of the system, from initial concepts to a robust, event-driven pipeline, highlighting the practical challenges encountered during deployment. A significant portion of this paper is dedicated to a case study on the debugging process, covering common software errors, platform-specific permission issues, and rare, environment-level platform bugs. The final architecture operates at a near-zero cost, demonstrating a viable model for individuals to build sophisticated AI-powered financial tools. The operational application is publicly accessible, and the complete source code is available for review. We conclude by discussing the role of LLMs in financial analysis, the importance of robust debugging methodologies, and the emerging paradigm of human-AI collaboration in software development.