One figure
One idea, drawn once
Every lesson is built around a single drawing. Read the notes and you are looking at the same figure the video animates, so nothing is lost moving between them.
Skillmaxing
Python with AI · Inference engineering · System design
303 lessons · 131 narrated videos · 10 free, start to finish.
Python with AI · Inference engineering · System design
See how agents work. Then build it.
Narrated, animated lessons and written notes for building AI backends in Python and designing systems that scale, with one real project running through every course.
10 people learning right now
03.01 CONCURRENCY VS PARALLELISM
Figure 2Three requests on one thread. The blue is work; everything pale is the request waiting on someone else.
303
lessons
131
narrated videos
222
animated figures
10
free, start to finish
Courses
Three tracks, one account
All three are written the same way and share one account. The free lessons are complete lessons, not previews.
python with ai
Builder path
Python with AI
- Python 3.14, FastAPI, Pydantic
- Agents SDK, LangGraph, MCP
- Langfuse tracing and deployment
Open course
ai research
Research path
AI research
- Math foundations to LLMs
- Landmark papers, in order
- GPU and tooling practice
Open course
system design
Engineering path
System design in depth
- Notes and case studies
- From-scratch implementations
- Built for senior interviews
Open course
inference
engineering
Production path
Inference engineering
- GPUs, KV cache and batching math
- Quantization, speculation, parallelism
- Autoscaling and production serving
Open course
Inside every lesson
Watch · Read · Build
Every lesson is made three ways
The video, the notes and the exercise use the same drawing, so nothing is lost moving between them.
- 01
Watch
Four minutes, narrated, with the diagram changing as the explanation moves.
- 02
Read
The same figures in writing, numbered and captioned, honest about what breaks.
- 03
Build
A Try it that adds one working piece to the project you are carrying.
Inside Skillmaxing
What an account actually gets you
Six things you use every week, drawn the way they actually work.
Watch · Read · Build
Three ways through the same thing
Four narrated minutes, the written version with numbered captions, and a Try it that adds one working piece to the project you are carrying.
Reading order
36 papers, in sequence
Landmark papers ordered so each one prepares you for the next, each with a brief, the key ideas, and a question to keep in mind while you read.
One a day
A brief every morning
Ten focused minutes on one paper, summarised and placed in context. Mark it read and the calendar fills in behind you.
Your dashboard
Everything you finish, counted
Mark a lesson understood and it lands on your dashboard with the date, the streak, and the next lesson waiting in that course.
Interactive
Change a number, watch it react
Move one control — a cache size, a fan-out, a timeout — and watch the latency curve answer. The numbers are the lesson.
Beyond the courses
More ways to learn on skillmaxing.
01Interactive tools
LabsChange a number, watch the system react
02One paper a day
DailyA short brief every morning
03Who builds what
CompaniesThe teams behind the tools
04Curated shelf
ResourcesDocs, books and talks
05Where to start
RoadmapThe order that makes sense
06Reading order
PapersLandmark papers, in sequence
07What engineers use
ToolsThe stack, explained
08Long reads
BlogDeeper write-ups
Plans
One account, every course
Every course on every plan
New lessons every week
Local pricing at checkout
$10.50 /month
$10.50 today · then $15/month
Every current course with the shortest commitment.
Choose Monthly$5.78 /month
$69.30 today · then $99/year
The same full access, paid once a year.
Choose AnnualOne flat price
$50 paid once
One payment, no renewal
All current tracks, every future course, and every update.
See lifetime offerFAQ
Questions people ask before joining.
Everything worth knowing before you pick a path. If it is not here, ask and it will be added.
Where should I start?
Start free. If you want to build AI backends in Python, open Python with AI. If you want architecture judgment and interview practice, open System Design. If you want papers and models, open AI Research.
What is actually inside AI Research?
Twelve modules that move from math foundations to modern LLM practice, with landmark papers sequenced so each one builds on the last.
What is actually inside Python with AI?
Fifteen modules that take you from Python itself to a production AI backend: FastAPI and Pydantic, databases and real-time transport, calling OpenAI and Anthropic models, agents with the OpenAI Agents SDK and LangGraph, MCP servers, Langfuse tracing, and deployment, all built around one running project.
What is actually inside System Design?
Thirty modules of notes, design walkthroughs, company cases, and from-scratch implementations, ending with an interview module and a capstone.
Is this only for interview prep?
No. Interviews are one module. The rest is written for the work itself: designing, operating, and changing real systems.
Can I use it before paying?
Yes. Every course opens with free lessons, and the free ones are complete lessons rather than teasers.
How often does the material change?
Weekly. New lessons, corrections, and rewrites land continuously, and every plan includes them.
Start free
$10.50 for the first month
Then $15 a month, every course included. Cancel from your billing portal.
Or $50 once, for everything published now and later.