Autonomous Agents and Sequential Decision Making in AI

computer science

credits
4
recorded trimesters
T1
open to
year 3, 4
last offered
2026 T1

about this course

This course examines how autonomous computational systems observe, reason, learn, and act across increasingly complex environments. Structured around a unified “agentic systems” framework, the course introduces students to sequential decision-making through deterministic search, strategic planning, probabilistic reasoning, reinforcement learning, and modern tool-using AI agents. Rather than treating classical AI, reinforcement learning, and contemporary LLM-based systems as disconnected paradigms, the course presents them as successive architectural layers of intelligent agency. Students progressively construct computational agents that evolve from search-based solvers into adaptive systems capable of planning under uncertainty, learning from interaction, invoking external tools, maintaining memory, and operating within safety constraints. Through an integrated practical lab structure, students maintain a cumulative software repository that is incrementally extended throughout the semester. By the end of the course, students will have designed and deployed a robust autonomous system that combines formal algorithmic rigor with contemporary AI engineering practices. The course emphasizes durable conceptual architectures over transient frameworks, ensuring students understand the computational foundations underlying modern AI systems while remaining adaptable to future technological developments.
in these programmes2026 batch
  • computer science

    elective · 4-year major, 4-year double major, 3-year major, 3-year double major, 4-year minor, 3-year minor

planning & teaching

before this1
recently taught by1

Recorded in 2026 T1. Instructors may change.

Course data from Krea's term sheets (July 2026); programme rules from the departments' FYUP documents. Check anything that matters with your department.