UST :: 2026-1 :: Mandatory General Course

Phase 1: Onboarding & Literacy

WEEK 1 Phase 1

Orientation: Agentic AI as R&D Collaborators

Lecture: Course Overview: Shift to "Directing Agents"
Practice: Environment Setup: AI Studio, Antigravity
Discussion: Is AI a research assistant or a crutch?

Paradigm Shift Environment Setup

WEEK 2 Phase 1

Understanding Large Language Models

Lecture: The Brain of the Agent: Capabilities & Security
Practice: API Connection & Setting up Ollama
Discussion: The "Stochastic Parrot" Problem

Ollama API

WEEK 3 Phase 1

Structured Directing via Prompt Engineering

Lecture: The Art of Instruction: System Prompts
Practice: CLI Agent: Building a command-line interface agent
Discussion: Managing Expectations of AI capabilities

System Prompts Persona

WEEK 4 Phase 1

AI Orchestration based on Tools

Lecture: Giving Hands to AI: Understanding Function Calling
Practice: Custom Tools: Connecting Python functions
Discussion: The Director's Role: Human's contribution

Function Calling Tools

Phase 2: Workflow Automation & Design

WEEK 5 Phase 2

Human-AI Agent Interaction Design

Lecture: UX for Research: How to design interfaces
Practice: Web UI: Deployable agent app using Streamlit
Discussion: Core Competency: The "irreplaceable" skill.

Web UI Streamlit

WEEK 6 Phase 2

Automation of Data Input

Lecture: Automating literature review & data collection
Practice: Auto-Scraper: Extract meta-data without reading
Discussion: The Risk of "Not Reading": Maintaining deep insight

Scraper Metadata

WEEK 7 Phase 2

Automation of Data Output & Discovery

Lecture: Drafting reports and visualizing results
Practice: Auto-Drafting: Converting raw CSV data
Discussion: Authorship & Ethics: Flawed hypotheses responsibility?

Auto-Drafting Visualization

WEEK 8 Mid-term

Mid-term Evaluation: Proposal

Lecture: Proposal of Agentic AI System
Practice: Video Pitch (3 to 5-minute pre-recorded video)
Discussion: Peer-Review: Assessing feasibility of colleagues

Mid-term Proposal

Phase 3: Management & Reliability

WEEK 9 Phase 3

Design of Task and Workflow

Lecture: Algorithmic Thinking: Designing robust workflows
Practice: Flow Architecture: Multi-step research logic
Discussion: Ambiguity vs. Clarity: Why agents fail at vague tasks

LangGraph Workflow

WEEK 10 Phase 3

Evaluation of Output from Agent

Lecture: Quality Control: Metrics for AI performance
Practice: Evaluation System: Designing rubrics
Discussion: The Consensus Dilemma: AI vs Human experts?

Metrics Evaluation

WEEK 11 Phase 3

Self-feedback Loop Agent System

Lecture: Metacognition: Making agents fix their errors
Practice: Self-Correction Pipeline: "Reflection" node
Discussion: Nurturing AI: Mentorship for AI systems?

Reflection Self-Correction

WEEK 12 Phase 3

Memory and Retrieval-Augmented Gen

Lecture: Long-term Context: Vector databases
Practice: RAG Pipeline: Remembering past papers
Discussion: Selective Memory: Remembering failed experiments?

Vector DB RAG

Phase 4: Collaboration & Future

WEEK 13 Phase 4

Multi-agent System

Lecture: Team Dynamics: Orchestrating multiple roles
Practice: Role-Playing Simulation: 2-agent debate
Discussion: Leadership: Managing a digital workforce

Dynamics Role-Playing

WEEK 14 Phase 4

Human in the Loop (HITL)

Lecture: Safety & Audit: Designing "Stop" buttons
Practice: Audit Interface: Human approval step
Discussion: The Gatekeeper: Filtering biased agents

Safety HITL

WEEK 15 Phase 4

Standardization & Expansion

Lecture: Connectivity: Model Context Protocol (MCP)
Practice: Interoperability: Connecting external agents
Discussion: Sustainability: Ethical research ecosystem

MCP Interoperability

WEEK 16 Final

Final Showcase: Demo Day

Lecture: Live Demonstration: Agentic AI System
Practice: The Director's Report: Reflection
Discussion: Peer-Review: Assessing AI completeness

Demo Report