Learning Outcomes
Upon completion students will be able to:
- Distinguish business workflows, AI-augmented workflows and autonomous AI agents.
- Combine orchestration, retrieval and RL frameworks for finance.
- Implement multi-agent crews that plan, code and self-correct using OpenAI Re- sponses,
- AutoGen and CrewAI.
- Buildknowledge-integratedagentswithCorrective
- RAGandthought-tracinghooks.
- Augment LLM agents with RL reward layers for tool selection, compliance and performance.
- Evaluate agent safety with AgentBench tasks, tail-risk metrics and audit trails.
Prerequisites
- Working knowledge of Python programming.
- Introductory understanding of financial markets and instruments.
- Basic familiarity with machine-learning concepts; prior RL is helpful but not required
Assessment Methods
- Practical Labs (60%) – graded coding exercises.
- Final Project (30%) – finance-sector agent meeting safety & performance targets.
- Participation (10%) – engagement and peer reviews.
Recommended Resources
- Frameworks: OpenAI Responses, AutoGen, CrewAI, LlamaIndex, LangChain.
- RL & Tool-use: ARTIST, ReCall, verl.
- Benchmarks/Simulators: AgentBench, TradingAgents, FinRL-Meta, SWE-agent.
- Key Papers: Tree-of-Thoughts, Algorithm-of-Thoughts, CRITIC, Corrective-RAG, Turn-Level
Credit Assignment, Self-Taught Evaluator. - Interpretability: Anthropic Circuits, Thought-Tracing Toolkit.
Download Event PDF