Building with LLMs

One-liner

Study LLMs in a small group of motivated founders & PMs, and figure out what you can build with them for your startup/business (e.g., add AI features, build internal tools, launch a new product)

Course stats (Oct 2025)

• 3 years
• 11 cohorts
• 170 students (YC founders, PMs from cool companies like Deel & Google)
• solid outcomes and lots of positive reviews
• 100% completion rates for c9 & c10

What it is

A 5-week online course aimed at:

1. understanding LLMs deeply
2. figuring out what you can build with them for your startup/business (e.g., internal tools, AI features, new products)
3. learning the basics of building so that you could prototype a v1 yourself and then hand off to engineers/cto

This course is designed for non-technical founders & PMs. That being said, ~20% of our students so far have been technical and they all enjoyed it too. If you’re technical and uncertain whether you should take this class or not, email us and we’ll help you decide.

How it works

• small study group of 7-12 founders/PMs
• structured, well-researched curriculum, updated as of Oct 2025
• two hands-on instructors, both are YC founders
• weekly speakers like:
  • Gordon Wintrob (co-founder and CTO of Newfront, $2.2bn insurance tech co)
  • Vaibhav Gupta (co-founder of Boundary YC W23, ex-Google, ex-Microsoft)
• SOTA learning practices like active recall, spaced repetition, metacognition, etc.
• assignments and group calls to keep you accountable
• course project!

Reviews

see more reviews here

Outcomes

see more outcomes here

Syllabus

Basically we go over the last 5 years of LLM evolution and build up your understanding gradually, from base models like gpt3 to assistant models (+tools +multimodality) to reasoning models & 2025 SOTA (+agents).

See the full syllabus here.

In the last two weeks of the class, you build a project, which usually is some piece of software that leverages LLMs (e.g., internal tool, AI feature, new product prototype). See some examples in the outcomes section above.

Logistics

• fully remote
• 5 weeks, ~1-2h a day to make sure it really sinks in and make progress on projects
• 3 weeks of theory, 2 weeks of project work
• 60% async (reading, watching, tinkering, assignments, etc.) and 40% live (speaker lectures, group discussions)
• live lectures and group calls are on Thursdays and Fridays at 9 am PT

Note: you really need to allocate at least 1h per day to the course, otherwise it won’t work. As I like saying, “1h a day or stuff fades away.” Karpathy explains why in this twitter thread.

Instructors

#1 Vasili Shynkarenka

• mostly worked on chatbots and voice assistants, since 2015
• built a software agency BotCube in ~2016-2018, creating chatbots for large enterprises like $2bn/year retailers
• YC founder, Storyline W18 (no-code Alexa skills builder)
• teaching LLMs for 5 years now
• focus: LLM fundamentals, LLM applications
• linkedin, AI blog

#2 Charlie Guo

• AI engineer at Pulley W20 (equity mgmt for startups)
• prev co-founder & CTO at Crowdmade S13 (merch creation and fulfillment for creators)
• writes AI newsletter w 20k subs called Artificial Ignorance
• Stanford CS, technical
• focus: AI engineering
• linkedin

Next cohort details (12th)

Cohort #12 will run from Nov 10 to Dec 15. It will the last cohort of this year.

Pricing

The course costs $3,000, with a 20% discount if you buy multiple seats.

How to sign up

Fill in this form and we’ll be in touch in ~24h.

PS Vibes check

It has been repeatedly reported that our courses have really good vibes!

See some pictures from the past few batches below:

9th cohort kickoff! featuring Jenna Blaicher-Brown (exited Shipamax W17 to $40B+ WiseTech), Matthew Busel (co-founder of Whalesync S21, now building agents at Basis), and others

10th cohort kickoff! featuring Zach Sims (Codecademy Founder/CEO, exited for $525m to Skillsoft), Andrey Artemenko (Carrefour VP of ecommerce), and others

Gordon Wintrob (co-founder and CTO of Newfront, $2.2bn insurance tech co) stopping by to talk about how they use LLMs internally and to improve their product

Elena Samuylova of Evidently AI S21 giving a lecture on evals; Charlie Guo (course instructor) digging in and asking re vibe evals

Vasili Shynkarenka (course instructor) explaining how LLMs learn by having Chinese-naive students try and make sense of Chinese writing; read more here

Vasili & Charlie as course instructors answering a (very long!) list of Mike Mahkow's open questions about LLMs

Jon Dahl of Mux W16 pitching his course project - n8n pipeline for story generation; Nazli Danis of Paloma S25 giving feedback!

THE END