简体中文
EnglishWhat Is i am trask?
Updated
i am trask (iamtrask.github.io) is a personal machine learning craftsmanship blog, not a product or service site. It publishes tutorials and essays on AI and deep learning, with recent posts covering decentralized AI, encrypted deep learning, and training methods that depart from standard backpropagation. If you are looking for a hosted tool, an API, or a commercial offering, this is not that; if you want long-form technical writing and walkthroughs on applied machine learning, it is worth reading.
What the site is
The site describes itself as "A Machine Learning Craftsmanship Blog." Its structure is a standard blog: a home page listing posts with dates and authors, plus Home, About, and Contact sections. Posts are attributed to iamtrask and carry publication dates, and the archive extends back through "Older Posts."
The emphasis on "craftsmanship" fits the content: the posts are written as tutorials and reflective essays rather than product announcements or news summaries.
What kind of content it publishes
The visible posts fall into two broad groups.
Tutorials with runnable code. These walk through an implementation and explain the mechanics along the way:
- Decentralized AI in 50 Lines of Python — a peer-to-peer AI that answers your friends' messages using local data and privacy controls (April 7, 2026).
- Deep Learning without Backpropagation — a tutorial on DeepMind's Synthetic Gradients (March 21, 2017).
- Building Safe A.I. — a tutorial for encrypted deep learning (March 17, 2017).
Essays and commentary. These are shorter pieces of opinion or analysis:
- Is the Universe Random? — thoughts on unpredictability in an increasingly predictable world (June 19, 2017).
- Safe Crime Detection — homomorphic encryption and deep learning for what the post frames as more effective, less intrusive digital surveillance (June 5, 2017).
The recurring themes are privacy-preserving machine learning (encrypted computation, local data), alternative training methods, and the social implications of AI systems.
Who it is for
| Reader goal | Is this site a fit? |
|---|---|
| Learn how a specific technique is implemented, step by step | Yes — the tutorials are written to be followed |
| Find a library, hosted service, or API to use | No — it is a blog, not a product site |
| Read opinion on AI and privacy trade-offs | Yes — several posts are essays on exactly this |
| Get current news or benchmark comparisons | Not primarily — posts are dated and topic-driven |
How to use it
Start from the home page and scan the post list; each entry gives a title, a one-line description, and a date, which is usually enough to decide whether a post matches your interest. Open the tutorial posts if you want implementation detail, and the essay posts if you want argument and context. The About and Contact sections cover the author and how to reach him.
One practical note: the archive is sparse and irregular. The visible posts jump from 2026 back to a cluster in 2017, so treat the blog as a set of standalone pieces rather than a continuously updated course. Check the date on any post before relying on its code or framing as current practice.
What it is not
There are no pricing pages, payment links, or sign-up flows on the site, and nothing in the visible content indicates a paid tier or account requirement. It presents itself purely as writing. If you need support, documentation, or a service-level commitment, look elsewhere.