What Is the Jingles (Hong Jing) Website About?

jinglescode.github.io is the personal site of Hong Jing, who writes under the name Jingles. It is a blog and portfolio focused on machine learning and neuroscience, written from the perspective of a data scientist and Alibaba PhD student. The site is informational rather than commercial: it presents articles, projects, and published work, and links out to Medium, LinkedIn, and GitHub. If you are looking for tutorials, conceptual explainers, or examples of applied ML work, the site is aimed at that kind of reader.

Who runs the site

The site identifies its author as Hong Jing (Jingles), described as a data scientist and Alibaba PhD student. The writing sits at the intersection of two fields: machine learning methods and neuroscience questions. That combination shapes the content — posts tend to explain a technique and then connect it to how brains or biological signals are studied, or the reverse.

What the site covers

The homepage lists articles across several recurring themes:

• Deep learning fundamentals — for example, how convolutional layers work in neural networks, and an illustrated guide to Transformers.
• Language models and sequence problems — state-of-the-art language models in 2020, and the three types of sequence prediction problems.
• Reinforcement learning — coordinating multiple RL agents on Overcooked.
• Applied computer vision — introducing Objectron as an "ImageNet" for 3D object understanding, and how radiologists used computer vision to diagnose COVID-19.
• AI and neuroscience — the relationship between AI and neuroscience, and decoding brain signals with machine learning and neuroscience.
• Practical and career-oriented posts — seven ways to get started in ML ranked from easiest to hardest, and setting up LaTeX in the Atom editor.
• Broader essays — what 200,000 years of humanity taught us about data storytelling, and whether GANs can be used beyond generating art.

The mix is deliberate: some posts are hands-on (tool setup, getting started), some are conceptual (optimizers, convolution), and some are commentary on where the field is heading.

How the site is organized

The navigation reflects a personal portfolio structure rather than a publication:

There is also a tag-based browse view ("Browse articles by tags"), which is the fastest way to find posts on a specific topic if you already know what you want.

When it is useful to you

The site fits a few specific situations:

• You want an explainer on a core ML concept — the convolution and Transformer posts are written as introductions, so they suit readers who know some ML but want the mechanism laid out.
• You are curious about the ML–neuroscience overlap — the brain-signal decoding and AI-and-neuroscience posts are the distinctive part of the site and are harder to find elsewhere in this form.
• You want a starting path into ML — the ranked "7 ways to get started" post is explicitly ordered by difficulty.
• You want to see applied examples — the COVID-19 computer vision and Objectron posts show techniques in a real problem context rather than in isolation.

It is less suited to readers who need reference documentation, current benchmark numbers, or maintained code libraries. The posts are dated (several reference 2020), so treat model comparisons and tooling advice as historical context and check current sources before relying on specifics.

A practical way to use it

Start with the tag browse page if you have a topic in mind; otherwise scan the blog index and pick by theme. For conceptual gaps, read the fundamentals posts first (convolution, Transformers, sequence problems), then move to the applied pieces. If you want to follow the author's work beyond the site, the Medium, LinkedIn, and GitHub links in the navigation are the intended next step — GitHub in particular is where code for project posts would live.