This repository contains information about deliverables, project ideas, and all things related to the final project.
DSAN 6725 is an applied AI course. The focus is on building production-quality AI agent systems that solve real problems. All projects must implement AI agents unless an alternative approach is explicitly approved by the professor.
This repo provides project ideas but you are not limited to these. You can explore other ideas but it is incumbent upon you (your team) to discuss these ideas and get approval before proceeding. The ideas suggested here are practically useful and have been selected because they have a reasonable chance of success in a 6-week timeframe.
This repo is organized into the following parts:
These project ideas have well-defined problem statements, but the implementation details are flexible and open to creative solutions. Use the descriptions provided as springboards for your own approaches to solving these problems.
- AI Tech Debt Forge - Multi-agent system for codebase modernization with persona-based validation
- AI Trading Strategist - Paper trading system using Alpaca API for strategy development and backtesting
- AI Spark Optimizer - Intelligent Spark job analysis with interpretable performance recommendations
- Cloud Cost Refinery - AWS cost optimization agent that generates executable cleanup commands
- AI Schema Harmonizer - Normalize schemas across SaaS tools with observability data focus
- AI Data Prep Pipeline - Universal document processing for RAG and vector search
Previous semester project ideas are available in spring-2025/.
All deliverables are described in deliverables.md.
Summary of what you will produce:
- Project paper (8-12 pages, conference format)
- Code repository (production quality)
- Working demo
- Presentation slides
- Conference-style poster
Can I do this project alone or with more than 4 people? No. Teams must have 2-4 members.
Can I use any model provider (OpenAI, Anthropic, Google, etc.)? Yes. Use whatever works best for your project.
What if I want to propose a different project idea? Discuss with the professor before Milestone 1. Your proposal should have a clear problem statement, feasible scope for 6 weeks, and an AI agent architecture.
- Course bookmarks repository
- LangChain, LlamaIndex, Claude Agent SDK