Uploader

The uploader's exact words: in the future, people without a programming code background can also possess such programming ability.

The uploader used AI-assisted development on a high-performance video editor project with 30,000 lines of Rust code, taking three months, at 12 times the speed of previous handwritten code.

The uploader developed a card-based tower-pushing game project entirely with AI, producing 46,000 lines of code in 9 days, averaging 5,000 net new lines per day, at 15 times the efficiency of the semi-AI mode.

The Agents.md handwritten by the uploader is only 22 lines (just over a dozen lines including spaces), while AI-generated versions tend to balloon to 80 lines.

The UP host believes that most AI usage tutorials online, especially paid courses, are essentially ripping people off, because large models are already capable enough and many demonstrations are just for show.

The creator's core judgment is: viral hits are not learned, they are torn down; tear down enough and use the formula skillfully enough, and you can make your own viral hit.

After finding the 3D route difficult, the uploader decided to switch to a 2D direction and continue exploring.

The uploader said they are not very good at modeling, and that the model's structure is unreasonable and rigging the movements is difficult.

Automotive testing is the low-threshold direction the uploader strongly recommends; his team recently sent hundreds of trainees into the automotive testing field, all of whom achieved employment.

The UP主 had previously generated a 2880×1920 version with another workflow, but it had three problems: no deinterlacing, only 416 seconds of runtime, and a bitrate as high as 120Mbps.

The uploader suggests not buying textbooks, because the textbook is already outdated from the day you place the order.

The UP host developed a vulnerability discovery Skill for the EDU education platform, whose skill.md defines a complete workflow: clarify objectives → asset collection → liveness probing and fingerprinting → attempt entry breakthroughs → vulnerability probing.

At the beginning of the tutorial, the UP host reminds: never use the techniques you learn to do anything illegal.

The uploader said in the video: "If I make this course sound simple, that actually means the practice part you do afterward will be very complex."

The course requires learners to deploy a "support library project" to a server, achieving a real deployment step rather than just running it locally.

The uploader provides a project list, and each student chooses two projects for hands-on practice.

The uploader tested 11 vision models with a self-made "Milk Frog Bench."

According to the UP host's own account, after studying at this pace for three months, companies began proactively approaching him to discuss opportunities.

The Bilibili creator ultimately landed a 30K large-model position through systematic learning and hands-on project experience.

The uploader is quite optimistic about the progress, saying "very fast, especially if you use Vibe Coding, it's not impossible to finish both in a day".