Skip to content

LLM Bootcamp

🚀 Announcing Full Stack Deep Learning LLM Bootcamp 🚀

  • In person in San Francisco on April 21-22
  • Learn best practices and tools for building LLM-powered apps
  • Network with a couple hundred other builders
  • Get cloud credits from OpenAI and other sponsors
Venn diagram showing that FSDL is at the intersection of a course, a hackathon, and a conference.

Why

We are at the cusp of a technology unlock of a magnitude not seen since the early days of the Internet. Rapid AI progress is fulfilling the old promise of Language User Interfaces (LUIs), which will revolutionize software just as GUIs did in the 90's.

The way AI-powered apps are built is rapidly changing:

  • Before LLMs, an idea would bottleneck on training models from scratch, and then it'd bottleneck again on scalable deployment.
  • Now an MVP based on pretrained, promptable LLM models and APIs can be configured and serving users in an hour.

An entirely new ecosystem of tools and tool vendors has formed around LLMs and LUIs. Even ML veterans are scrambling to orient themselves to what is now possible and figure out the most productive techniques and tools.

Engineers are asking themselves:

  • Are there any good open-source LLMs?
  • What is my moat if I rely on OpenAI APIs?
  • Is Prompt Engineering some kind of sick joke?
  • How can I gather and use feedback from users?
  • Should I be able to code a Transformer from scratch?
  • How exactly am I supposed to test these damn things?

What

We have put together a two-day program based on emerging best practices in the LLM community and the latest research results to help you answer these questions and make the transition to building.

Our goal is to get you 100% ready to build and deploy LLM applications and 100% caught up with the state-of-the-art.

What do I need to know already?

We aim to get anyone with experience programming in Python ready to start building applications that use LLMs.

Experience with at least one of machine learning, frontend, or backend will be very helpful.

And if none of the questions above resonate, this probably isn't for you.

Tentative Schedule

Friday (April 21)
9 am Registration & Breakfast
10 am 🚀 Launch an LLM App in 1 Hour
  • We'll go from idea to user-ready website in one hour
11 am 🗿 Foundations of Foundation Models
  • Learn the core concepts behind transformer architectures, self-supervised learning, and text generation
  • Develop clear intuitions for model internals based on the latest work in "reverse engineering" LLMs
12 pm Lunch and networking
1 pm ✨ Learn to Spell: Prompt Engineering and Other Magic
  • Learn how vendors like OpenAI, Cohere, and AI21 compare with each other and with OSS options like FLAN-T5 and GLM.
  • Prompt engineering tips and tricks: few-shot examples, chain-of-thought, formatting
  • Context engineering concepts: incorporating local information, wishlist-fulfillment architecture, long-term memory
  • Software tools: LangChain, GPTIndex, Everyprompt, dust.tt, etc.
2 pm 👀 Search 2.0
  • Jointly embedding multiple types of data for multi-modal semantic search
  • Choosing between vector stores (e.g. FAISS, Milvus, Pinecone, Weaviate, Vespa)
3 pm Coffee and networking
345 pm 👷‍♂️ askFSDL Walkthrough
  • Detailed breakdown of a well-documented sample project demonstrating use of LLM APIs and frameworks, traditional and vector databases, and user feedback ingestion
430 pm 🎤 Invited Talk from
Richard Socher: CEO and co-founder of you.com
515 pm 🎤 Invited Talk from
Peter Welinder: Director of Product at OpenAI
Saturday (April 22)
9 am Breakfast
10 am 🌳 Demo Garden
  • Present your cool project
  • Review other cool projects
11 am 🤷 UX for LUIs
  • Review of the best AI-powered apps today
  • Principles of successful design for AI-powered apps
12 pm Lunch and networking
1 pm 🏎️ Deploying and Learning in Production
  • Deploying on CPUs vs GPUs vs API-only
  • How to monitor models, trace chains, and record feedback
  • Methods for learning from users, like RLHF, and from chains of LLMs, like Constitutional AI
2 pm 🔮 Future Directions
  • Lightning tour of things that are surprisingly possible today
  • Building future-proof applications: what's around the corner
  • What are still hard research problems
3 pm Coffee and networking
4 pm 🎤 Invited Talk from
Harrison Chase: Creator of LangChain
4:45 pm 🎤 Invited Talk from
Reza Shabani: Training LLMs at repl.it
5:30 pm Discussion Panel: Building a Defensible Business

Who

We are Full Stack Deep Learning. We're a team of UC Berkeley PhD alumni with years of industry experience who are passionate about teaching people how to make deep neural networks work in the real world.

Since 2018, we have taught in-person bootcamps, online multi-week cohorts, and official semester-long courses at top universities.

As former academics, we always make sure that all of our materials become accessible for free, right here on this website.

Group photo of the attendees of FSDL March 2019 bootcamp Group photo of the attendees of FSDL August 2018 bootcamp Group photo of the attendees of FSDL November 2019 bootcamp

Instructor Team

Photo of Charles Frye
Charles Frye educates people in AI. He has worked on AI/ML tooling with Weights & Biases and Gantry since getting a PhD in Theoretical Neuroscience at UC Berkeley.
Photo of Sergey Karayev
Sergey Karayev builds AI-powered products as Co-founder of Volition. He co-founded Gradescope after getting a PhD in AI at UC Berkeley.
Photo of Josh Tobin
Josh Tobin builds tooling for AI products as Co-founder and CEO of Gantry. He worked as a Research Scientist at OpenAI and received a PhD in AI at UC Berkeley.

Invited Talks

Photo of Peter Welinder
Peter Welinder is VP of Product and Partnerships at OpenAI.
Photo of Harrison Chase
Harrison Chase is the creator of LangChain.
Photo of Richard Socher
Richard Socher is the co-founder and CEO of you.com.
Photo of Reza Shabani
Reza Shabani trains LLMs at repl.it.

When and Where

The event will be in-person and run all day on Friday, April 21, 2023 and Saturday, April 22, 2023 at the South San Francisco Conference Center.

Register

Regular
$950
Academic
$450

If you are a current full-time student or post-doc, enjoy this lower price.

Get discount code

Need to expense the registration costs? Check out our quick guide to reimbursement, including an LLM-generated email template for getting manager approval.

If you have any questions about the bootcamp, contact admin @ fullstackdeeplearning.com. We cannot honor requests for additional discounts.

Can't make it?

Join 24K subscribers and we'll notify you when we eventually put the materials online.

While you're at it, follow us on Twitter and YouTube !

Sponsors

We're grateful to a number of sponsors who are helping us make this event happen.

Compute Credit Sponsors

We're partnering with compute vendors to offer free cloud credits to all attendees, including $500 in credits for data-centric compute infrastructure on Modal, $50 in credits to query LLMs via the OpenAI API, $50 in credits for running serverless GPU jobs on banana.dev, $50 in credits to run servers in the LambdaLabs GPU Cloud, and 1000 cycles on repl.it.

Direct Sponsors

We are deeply grateful to all of the sponsors who helped us make this event more affordable for attendees.