Project Showcase

Outstanding projects from course participants. These projects demonstrate innovative applications of agentic AI concepts.

Want to feature your project? Submit it through Show & Tell Discussions or contact the instructor.

### Research Paper Analysis Agent {: .project-title }
Multi-Agent RAG LangGraph
**Team**: Anonymous (Spring 2025) An autonomous system that analyzes research papers using a team of specialized agents: - **Extractor Agent**: Pulls key claims, methods, and results - **Critic Agent**: Identifies limitations and potential issues - **Connector Agent**: Finds related papers using semantic search - **Summarizer Agent**: Generates structured summaries **Technologies**: LangGraph, ChromaDB, Claude 3.5 Sonnet **Key Results**: 85% agreement with human reviewers on paper quality assessment
View Demo GitHub Paper
--- ### Financial News Trading Agent {: .project-title }
ReAct Tool Use Verification
**Team**: Anonymous (Spring 2025) A verified news trading system with hallucination prevention: - Real-time news ingestion from multiple sources - Multi-step verification using Chain-of-Verification - Sentiment analysis with confidence calibration - Trade signal generation with explanation **Technologies**: OpenAI GPT-4o, FinancialModelingPrep API, Custom verification chain **Key Results**: 23% reduction in false positive trading signals
Architecture Results
--- ### Code Review Crew {: .project-title }
CrewAI Code Agent AST
**Team**: Anonymous (Spring 2025) Multi-agent code review system simulating a development team: - **Security Analyst**: Scans for vulnerabilities (OWASP top 10) - **Performance Expert**: Identifies optimization opportunities - **Style Reviewer**: Checks code conventions and readability - **Test Strategist**: Suggests test cases **Technologies**: CrewAI, Tree-sitter, CodeLlama **Key Results**: Identified 40% more issues than single-agent baseline
Demo Video GitHub

Project Categories

Autonomous Research Assistants

Project Description Agent Type
Literature Mapper Graph-based paper relationship visualization GraphRAG
Experiment Designer Proposes and validates research hypotheses Planning
Citation Checker Verifies claims against cited sources Verification

Domain-Specific Agents

Project Domain Key Innovation
Legal Contract Analyzer Law Clause extraction with precedent retrieval
Medical Triage Assistant Healthcare Symptom reasoning with safety guardrails
Supply Chain Optimizer Logistics Multi-objective planning under uncertainty

Tool-Augmented Systems

Project Tools Used Capabilities
Data Science Copilot Python, SQL, Pandas Natural language to analysis pipeline
Web Research Agent Browser, Search, APIs Autonomous information gathering
DevOps Automator Git, Docker, K8s Infrastructure as conversation

Submission Guidelines

Requirements

Your project should demonstrate mastery of course concepts:

  1. Agent Architecture: Clear design with appropriate agent pattern
  2. Tool Integration: At least 2 external tools or APIs
  3. Reasoning: Explicit reasoning trace (CoT, ReAct, etc.)
  4. Evaluation: Quantitative metrics or user study results
  5. Documentation: README, architecture diagram, examples

Evaluation Criteria

Criterion Weight Description
Innovation 25% Novel approach or application
Technical Depth 25% Appropriate use of course concepts
Implementation 20% Code quality and completeness
Evaluation 15% Rigor of testing and metrics
Presentation 15% Clarity of documentation and demo

How to Submit

  1. Prepare your repository:
    • Clean, documented code
    • README with setup instructions
    • Architecture diagram
    • Demo notebook or video
  2. Create submission:
    • Post in Show & Tell
    • Include: title, team, summary, links
    • Tag with relevant topics
  3. Presentation:
    • 10-minute demo during final weeks
    • Q&A with class and instructor

Past Semesters

Spring 2025 (Current)

Projects will be added as they are completed and approved for showcase.

Have a project to showcase?

Submit Your Project

Back to top