Factored
GenAI Fundamentals and Prompting
Understand how large language models work, from tokens, caching and sampling to agents, and learn to prompt them as an engineering discipline with structured outputs, prompt policy and A/B testing.
About This Course
GenAI Fundamentals and Prompting is the entry point to the GenAI track at Factored. It is for engineers who can write Python and call APIs but have little or no hands-on experience with large language models, and it moves them from using chat assistants casually to understanding how models process text and building prompts that hold up in production.
The course pairs free vendor courses from Microsoft, Anthropic and DeepLearning.AI with original deep dives written for Factored engineers. Each subsection opens with the vendor material as a guided introduction, and the deep dives then go further on the points that decide cost, latency, reliability and quality in real systems.
What You Will Learn
The first half builds an accurate mental model of how LLMs work. After a section on modern Python for AI systems, covering packaging, strict typing with Pydantic and asyncio, you learn what a foundation model is and how pre-training, instruction tuning and preference alignment produce an assistant. You learn how Byte-Pair Encoding tokenizers split text and why English usually costs fewer tokens than languages such as Spanish or Portuguese, how context windows are shared, why output tokens cost more than input tokens because of the prefill and decode phases, how the KV cache and prompt caching work and why static prompt content belongs before dynamic content, how temperature, top-p and top-k shape output, and why LLMs are non-deterministic even at temperature zero. The section closes with DeepLearning.AI's Agentic AI course and a deep dive on writing tools that agents can use.
The second half turns that understanding into prompting practice. You compare model providers and API interaction modes, including streaming and batch processing, then learn the fundamentals of prompting, the difference between system and user prompts, and techniques from zero-shot, few-shot and meta prompting to chain of thought and Tree of Thoughts. You learn context engineering inside a single prompt, output formatting and structured outputs with JSON Schema, and finally how to treat prompts as engineering artifacts through acceptance criteria and critique, a versioned prompt policy, and A/B testing. A graded lab applies the section end to end.
Who This Course Is For
The course is written for software engineers, data engineers and data analysts starting the GenAI track. It assumes you are comfortable with Python, REST APIs and Git. It does not assume any prior experience with LLMs, machine learning or prompt engineering; every concept is introduced before it is used or tested.
How the Course Is Structured
A Welcome section explains how the course works and how to request the API key the labs need. Modern Python for AI Systems covers packaging, typing and asyncio and ends with a lab. How LLMs Work has five subsections: What Generative AI Is; Tokens and Context; Inside Inference: Speed, Cost, and Caching; Sampling and Non-determinism; and From Models to Agents. LLM Prompt Systems and Structured Outputs has five subsections: Providers and API Interaction Modes; Prompting Fundamentals; Prompting Techniques; Context and Output Formatting; and Prompts as Engineering Artifacts, followed by the lab and a short course feedback form. Short knowledge checks appear only where a subsection teaches material you can apply.
Time Commitment and Completion
The course is self-paced and follows the program calendar for weeks zero and one of the GenAI track. The longest external item is DeepLearning.AI's Agentic AI course, which is best taken over several sittings. Progress is tracked automatically in the course outline as you view units, answer knowledge checks and push your lab work, and the labs are graded from your repository.