Viktor.comAn AI coworker that actually does the work
LiminalML
Six-stage study sessions for mastering ML & SWE
9 followers
Six-stage study sessions for mastering ML & SWE
9 followers
LiminalML runs every topic through the same six stages: intuition, the math step by step, runnable code, the questions an interviewer actually asks, and a retrieval drill so it sticks. Built on Claude. 157+ ML & SWE topics, with math, diagrams, and code rendered properly inline. Every session saves. Read the concept pages free. 20 free sessions/month.
The way it bundles intuition, math, code, and interview questions for each topic is genuinely useful, and the retrieval drill at the end actually helps things stick instead of fading by morning.
@safiyegmene97v Thanks for commenting, that last part is exactly why the format exists. The retrieval drill is the whole point as it's easy to read a good explanation, feel like you've got it, and have nothing left by morning (I did this to myself constantly). Forcing active recall at the end is what turns definitely helps things stick, and when you make an account and upload your resume, it'll reference that in the retrieval stage and will ask you questions where a certain topic might be relevant to one of your projects, which is a feature thats helped me the most.
Try it out!
The six-stage format actually helps me slow down on topics I usually skim past, especially the retrieval drill at the end. Wish the inline math rendered cleaner, but overall a solid way to revisit fundamentals without bouncing between tabs.
@nuriyepbrr Thanks for the feedback! The math renders cleaner in the sessions once you've signed in. The demo on the homepage doesn't render as well currently.
the six-stage structure is genuinely clever, especially the retrieval drill at the end. most study tools stop at explanations, but forcing yourself to recall right after learning is where concepts actually stick. solid execution on something that could have easily been shallow.
@niyazipv74 Appreciate that! "could have easily been shallow" was the exact thing I wanted to make sure wouldn't happen. It's easy to ship something that looks like a study tool but is really just reworded explanations or links to other resources constantly, I wanted everything to flow well in the sessions. So I knew the math had to actually be derived, the code can be modified in the practice tab, and the questions have to be ones a real interviewer would ask. The retrieval drill definitely helps with deeply understanding everything, and asking followup questions within each session is something I do often and that has really helped with mastering topics. The ML topics especially benefit from the six stages, as it's helpful having the math, visuals, and code all in one place.
What topics are you learning?