PAID ENROLLMENT OPENING SOON / PAYMENTS OFF / FREE WORK-READY BY APPROVED PARTNER CODE
Student and partner sign-in temporarily unavailable while email setup completes. Public previews and interest list remain available.
PUBLIC PREVIEW / DRAFT — not open for enrollment
d↗Digital Business
AcademyExplore the courses ↗
All courses / Advanced / Digital Business series
C1 / Advanced

Autonomous AI Agents: Architecture & Engineering

Explore agents as decision processes (mdps), llm fundamentals: transformers, tokenization, sampling, prompting, tool calling, structured output. Work through applied exercises and develop a project you can explain, test, and improve.

A certificate issued by this academy is a certificate of completion only. It is not accredited. It is not approved by the Ohio State Board of Career Colleges and Schools (SBCCS) or any other state agency. No income or employment outcome is promised.

12 modules≈149 learner hoursSelf-paced draft
YOUR TAKEAWAYS

What you will practice

  • Explain the core theory of agents as decision processes (mdps).
  • Apply methods from prompting, tool calling, structured output.
  • Analyze trade-offs in memory & retrieval.
  • Evaluate solutions involving evaluation & benchmarks.
  • Design and build deployment & observability.
  • Produce the capstone: deployed agent with eval suite + incident runbook, defended to a human reviewer.
THE LEARNING PLAN

Inside the course

  1. 01Agents as decision processes (MDPs)
  2. 02LLM fundamentals: transformers, tokenization, sampling
  3. 03Prompting, tool calling, structured output
  4. 04Planning & reasoning loops (ReAct, plan-execute)
  5. 05Memory & retrieval
  6. 06Multi-agent coordination
  7. 07Evaluation & benchmarks
  8. 08Safety, guardrails, approval matrices
  9. 09Cost/latency engineering
  10. 10Deployment & observability
  11. 11Agent security: prompt injection, sandboxing, and secrets
  12. 12Governance, human–agent operations, and capstone integration
TRY THE MATERIAL

A little of what is inside.

EXCERPT FROM THE STUDENT MATERIAL
The word agent is overloaded. Marketing pages use it for any chatbot with a logo; researchers use it for anything that maps observations to actions. This course needs a definition precise enough to engineer against. We will use the following: an agent is a system that repeatedly observes some part of an environment, chooses an action from a set of available actions, executes that action so that the environment changes, and does so in pursuit of an objective that it can be evaluated against. Every part of that sentence carries weight. "Repeatedly" separates an agent from a single-shot classifier or a one-turn question-answering model.…

Excerpt from C1. The full curriculum is not publicly downloadable in this preview.