Tech with Mak Maps LLM Post-Training

This is a practical mental-model checklist for engineers navigating modern model adaptation, though implementation experience and evaluation still matter more than memorizing technique names.

• –Separates pretraining from the behavioral and capability-shaping work that follows.
• –Covers parameter-efficient fine-tuning, making customization more accessible without retraining entire models.
• –Connects preference optimization and reinforcement learning to current reasoning-model workflows.
• –Complements hands-on material such as [DeepLearning.AI’s post-training course](https://www.deeplearning.ai/courses/post-training-of-llms), which applies SFT, DPO, and GRPO.
• –Best used as a study or interview framework, then validated through real datasets, reward design, and evals.