ROMEOADVANCED ACADEMY

1. Read three papers this week

The single best thing you can do this week is apply the three-pass strategy to three real papers. Not three deep reads — three Pass-1 triages, possibly one Pass-2 if any one paper interests you enough. The discipline is the point. By the end of the week you should have your three-sentence summaries written down and a feel for whether you want to do another week of it. After a month, you will know whether this is a sustainable habit for you.

2. Go deeper — Path C in the full programme

This free course is, deliberately, an introduction to the research-reading discipline cultivated through Path C — Applied AI Research, the Tier 3 specialisation of the Integrated AI Program. Path C is 60 ECTS across ten courses:

C1

Mathematics for AI Research

Linear algebra, probability, optimisation, information theory — the working maths a researcher needs daily.

C2

Research Methods and Reproducibility

Experimental design, statistical analysis, ablation discipline. The institutional craft of doing research that holds up.

C3

Frontier Methods in Deep Learning

What the state of the art is across vision, language, RL, and multi-modal — and how to read the frontier as it moves.

C4

Alignment and AI Safety

The conceptual frame and the technical methods. From RLHF to constitutional AI to scalable oversight.

C5

Mechanistic Interpretability

Opening the black box. Circuits, features, dictionary learning, the interpretability research programme.

C6

Reinforcement Learning Theory and Practice

From the Bellman equation to PPO to multi-agent RL. The mathematical heart and the engineering reality.

C7

Probabilistic and Bayesian Methods

Bayesian inference, latent-variable models, the role of uncertainty in serious modelling.

C8

Research Communication

Paper writing, conference talks, technical blogging. The output side of research.

C9

Open-Source Research Contribution

How to contribute to AI codebases, evaluation harnesses, and benchmarks. The collaborative infrastructure of the field.

C10

Applied Research Capstone

An original piece of research, mentor-supervised, ending in a paper, a code release, or a blog post that adds to the field.

Path C launches with the December 2026 cohort. Applications are open now. Apply or write to hello@romeoadvancedacademy.com.

3. Read these to go further

The shortest, most useful reading list for the research-reading craft.

  • "How to Read a Paper" by Srinivasan Keshav (2007). Two pages. Free online. The source of the three-pass technique. Required reading.
  • The Distill journal archives at distill.pub. Although the journal stopped publishing new articles in 2021, the archives remain the gold standard for accessible technical writing on AI.
  • Sebastian Raschka's reading notes at sebastianraschka.com. An example of what a personal reading-summary practice looks like in public, sustained over years.
  • The Anthropic research blog, the DeepMind blog, the OpenAI research page. Companion posts to recent papers. Often more readable than the papers themselves.
  • "The Bitter Lesson" by Rich Sutton (2019). A short blog post that frames much of modern AI research. Useful context for why the field has gone the way it has.

4. Try the other free courses

Free course · 5 lessons

Build Your First AI Agent

If you read about agents in a paper and want to build one yourself, start here. The fundamentals course.

Take this course

Free course · 5 lessons

AI Security Foundations

The security frame on the AI systems you read about. OWASP for LLMs, MITRE ATLAS, defence in depth.

Take this course

Free course · 5 lessons

The EU AI Act for Non-Lawyers

The regulatory lens on the systems the research you read leads to. Plain language. Real obligations.

Take this course

Tell us what we missed

This is the first version of this course. The research-reading craft is changing — AI assistants are increasingly part of it, the volume keeps rising, the venues evolve. Tell us what we got wrong, what we left out, what worked for you and what did not. Email hello@romeoadvancedacademy.com with "Paper-reading course feedback" in the subject. We revise.

The December 2026 cohort opens for applications.

Path C — Applied AI Research. Sixty ECTS, ten courses, mentor-supervised capstone, online. For learners pursuing original research or doctoral preparation.

Apply now