LearnAISign in
Phase 11 · 6 lessons · 19 h · about 1 week

🚀 Research, Projects & Career

Read the papers, ship the projects, steer the career

Lessons

11.1How to read (and reproduce) papersThree passes, one notebook, and the habit of re-deriving3 h11.2Landmark papers: the reading listTen papers that explain the field, and how to read them in order4 h11.3Portfolio project ladderFifteen projects, each with a 'done when', from tabular ML to a paper reproduction4 h11.4ML and LLM system design and interviewsA framework, two worked cases and the question families you will meet4 h11.5Staying currentA 30-minute daily routine that separates signal from hype1 h 30 min11.6Frontier 2026: where the field is headingSeven directions, the mechanisms behind them, and how to keep learning2 h 30 min

About this phase

You now know how models work from the dot product up to tool-using agents. This last phase turns that knowledge into three durable abilities. First, reading research: the field moves through papers, and the three-pass method plus a reproduction habit lets you keep up without a professor. Second, building a portfolio: fifteen projects, each mapped to a phase and each with an explicit "done when", so you can prove what you know with code, evals and demos rather than certificates. Third, steering a career: system-design thinking for ML and LLM products, the interview question families you will actually meet, a weekly routine for staying current without drowning in hype, and a sober map of where the field is heading in late 2026 so you can choose a lane deliberately.

By the end you can

  • Read a technical paper in three passes, extract its key equation and reproduce one of its results
  • Explain the contribution of the ten landmark papers that define modern AI and know what changed after each
  • Plan and publish portfolio projects with READMEs, evaluations and live demos that pass a hiring-manager's five-minute test
  • Walk through an ML or LLM system design from requirements to monitoring, and answer the common interview question families
  • Run a 30-minutes-a-day routine for staying current that filters hype from signal
  • Describe the late-2026 frontier (RL for reasoning, long-running agents, agent memory, world models, scientific AI, efficient inference, interpretability) and pick a direction to go deeper