Hi, I'm Mark!

To write an almanack one usually needs to be famous, successful, and old. At the moment of writing this sentence, I cannot offer you any of the three. Yet, I'm still doing it because it as good way to live an examined life, and it could help others to do the same.

I've come to believe there are only a few important timeless questions in life. Answering them well brings great rewards, and their insights are mostly true for everyone. To that end, I wanted to document my answers and share them with others. I do it for three reasons: to clarify my thinking, to invite criticism, and to help others.

My addition to the almanack genre is the medium itself. Writing online allows me to continuously refine my ideas. I’ll pose a timeless question, offer a timeless answer, and keep updating it as my understanding evolves. Hence, the evolving almanack.

Table of Contents:

0 — Introduction

1 — My Life Philosophy

2 — My Life Goals

3 — My Operating Principles

4 — What Problems to Work on (WIP)

5 — How to Solve Humanity Scale Problems (WIP)

6 — Field Evolution: Guesses and Selection (WIP)

7 — Great Day, Great Life (WIP)

8 — A Seamless Web of Deserved Trust

9 — Curiosity, Freedom, Progress, and Fallibilism

10 — Where I Come From

Appendix

Appendix 1: Questions

Appendix 2: Links

David's ideas are valuable but complex. People take years to understand them. This is a problem. And by then I knew that problems are soluble.

The solution? Handbooks simply explaining David's ideas written around chapter's main questions. This way you could explore ideas, and then challenge your understanding. This studying method is 50% more efficient than others like passive reading or highlighting (studies: 1, 2, 3, 4, 5).

I spent 800 hours re-explaining David's ideas in simple language and question-answer format, so you can understand them in 80. I also recorded two podcasts summarizing every chapter of both books. Lastly, with Elias Schlie we recorded a 3-hour podcast explaining main ideas in conversational format:

• The Beginning of Infinity Handbook ↗
• The Fabric of Reality Handbook ↗
• 4 months of David Deutsch in 3 hours ↗

(The handbooks first were realesed in Notion-doc format, I later repackaged them as Substack posts.)

Around 19 I realized that while I was ambitious, I had no idea what to direct my ambition to. It was also the last year of my business bachelor, so I ended up doing a Business Development internship (turns out that means sales) at Onramper as my first guess.

After 7 months I had a few learnings about “biz-dev at crypto startup” hypothesis. Mainly, that while sales is a crucial skill, I don't want to dedicate my life to it — it lacks intellectual depth and long uninterrupted work hours I yearn. Getting a glimpse of other jobs at Onramper, and having overread Paul Graham's essays, I desperately wanted to try coding as a craft.

The problem is that not only I didn't have time or money to do another bachelor, I also had little faith in traditional education. So the solution came easily: take a year off and self-educate, similar to how Scott Young finished an MIT CS bachelor in one year.

Only difference is that I also wanted to get a job after that year! And while MIT is great at teaching you CS theory, it's not as good for learning practical on-the-job skills. So I also incorporated frontier practical skills related to web-development (my craft guess at the time). (Turns out Udemy is the best source for that.)

By the end I spent 8 months full time learning CS fundamentals and web-dev skills. Here's the list of projects I did:

• CS50x, CS50web.
• Udemy HTML and CSS course.
• Udemy Complete JS Course: shipped JS-heavy apps using asynchronous flows, API integrations, and OOP-style architecture. Learned DSA, JS engine and runtime.
• Started but didn't finish Udemy React Course (Next.js, Redux...).
• Made past version of my website. More practice projects: 1, 2, 3, 4, 5. For more see my repo: learning_cs_and_web_dev.

You can read more about my Self-Education Year, or watch me reflecting on it here. (I spend first 4 writing David Deutsch Handbooks, and at the end of the year pivoted to ML and technical AI safety.)

After exploring CS and web-dev for 8 months I realized I want to work on important problems, not 'B2B SaaS'. First thing that came to mind was longevity, i.e. making disease, aging, and death optional. I did my thesis on longevity metascience, and spend 2 months of summer studying Campbell Biology 10h a day — it was awfully fulfilling.

I was fully committed to dedicating my life to making death optional, or die trying :) But then, in September 2025 I fully priced in the revolution AI is, and will go through in the next few decades. Mainly, I realized that ASI will come before my old age. It might not come in the next 2 or 5 years, but it will come in the next 50, and that is early enough.

So I pivoted to "working for superintelligence to go well". My current interest is in technical AI safety. I take self-education seriously, similar to how professional athletes treat their practices. I follow consistent deliberate practice, divided into theoretical and technical parts.

Here's the list of project I have done since Sep 2025:

• Went through most of Hands on ML with PyTorch chapters and exercises in depth (was my main starting learning source).
• Went through most of AIMA chapters, studying heavily first few, and ML, DL ones.
• Completed ML4Good bootcamp and AFFINE Seminar (WIP!*).
• Won 2,500$ in AAAA Hackathon with Elias Schlie in AAAA hackathon.
• Implemented transformer encoder-decoder for NMT. Prior, implemented RNN (GRU) architectures for character prediction, sentiment analysis, and neural machine translation.
• Completed BlueDot's AGI strategy, TAIS course, and TAIS project(WIP*); also Harvard's CS2881 course. Read AI Safety Atlas and Introduction to AI Safety, Ethics, and Society.
• For good examples of my theoretical practice see: this scrappy explanation of transformer; this not-so scrappy explanation of AI, computers, and explicit-implicit coding; or this explanation of all main parts of ML.

For more on my studying approach or completed projects see techical_ai_safety_mastery repo.