Syllabus Application
CS 58012
Agentic AI Systems: Architecture and Deployment
Faculty
Faculty of Engineering and Natural Sciences
Semester
Fall 2026-2027
Course
CS 58012 -
Agentic AI Systems: Architecture and Deployment
Time/Place
Level of course
Masters
Course Credits
SU Credit:3, ECTS:10
Prerequisites
CS 555 or CS 555
Corequisites
-
Course Type
Lecture
Instructor(s) Information
Utku Günay Acer
- Email: utku.acer@sabanciuniv.edu
Course Information
Catalog Course Description
This course offers an in-depth exploration of Agentic AI, focusing on the design and implementation of autonomous systems. Students will master essential concepts such as agentic workflows, multi-agent collaboration, and reasoning frameworks like ReAct. The curriculum emphasizes practical development, covering technical skills in tool integration, memory management, and context engineering. Beyond architecture, the course addresses operational challenges in LLMOps and AgentOps, preparing students to build, evaluate, and scale robust AI solutions.
Learning Outcomes:
| 1. | Students will design and implement agentic workflows by architecting autonomous systems and selecting the most effective orchestration patterns for specific automation tasks. |
|---|---|
| 2. | Students will integrate external tools and APIs into AI agents, enabling them to interact with the physical and digital world through function calling and the Model Context Protocol (MCP). |
| 3. | Students will apply advanced reasoning frameworks, such as the ReAct framework and context-aware planning, to enable agents to solve complex, non-linear problems. |
| 4. | Students will execute reflection and self-improvement cycles within agentic systems to programmatically refine output quality and ensure iterative performance gains without manual intervention. |
| 5. | Students will manage agent memory and context by optimizing retrieval strategies and implementing various memory types to work efficiently within Large Language Model context windows. |
| 6. | Students will evaluate and benchmark agent performance using industry-standard metrics and real-world testing protocols to provide quantitative evidence of system effectiveness. |
| 7. | Students will deploy and scale autonomous systems by navigating the challenges of LLMOps and AgentOps, including monitoring, error handling, and robust exception recovery in production environments. |
| 8. | Students will develop multi-agent collaboration protocols to coordinate communication and task-sharing between multiple specialized agents to solve large-scale problems. |
Course Objective
The objective of this course is to provide students with a comprehensive, hands-on foundation in the principles, architectures, and operational practices required to design, deploy, and scale autonomous Agentic AI systems.Students will bridge foundational Large Language Model capabilities with practical engineering—gaining the architectural mindset and technical expertise to develop resilient, tool-augmented, multi-agent solutions capable of solving complex real-world automation challenges.
-
Course Materials
Resources:
No text book is required. The students will be given occasional reading material
Technology Requirements:
A personal machine/server with enough resources to run a local LLM with reasoning capabilities