Live online · 10 Tuesdays

Agentic AI bootcamp

Learn to build AI agents that reason, plan, and act

From reasoning models to a multi-agent application you deploy and keep: ten weeks, 30 hours, three hours every Tuesday morning — so you build alongside the job you already have. Taught live online by practitioners who have built and run these systems in production at their own companies.

From $2,100 with early-bird pricing. Ten weeks, live online.

Next online cohort 202710 Tuesdays

Jan 5

10 weeks, Tuesdays 9 AM–12 PM PT · Live online

$2,100$3,000

30% off ends in

Days
--
Hours
--
Minutes
--
Who it's for

Who should attend

  • Engineers

    Software, backend, platform and solutions engineers

    Take an agent from notebook to production, and know what breaks once real traffic calls your tools.

  • Data scientists and ML engineers

    Data scientists, ML and AI engineers, analytics leads

    You already build models. Learn the layer around them: retrieval, orchestration, and evaluating outputs with no single right answer.

  • Product leaders and founders

    Product managers, product directors, technical founders

    Learn what an agent can automate and how to scope the work, well enough to tell a week from a quarter.

Curriculum

From reasoning models to production-ready agents

Every module pairs lecture with discussion and hands-on exercises. The program ends with a multi-agent application you build and keep.

One module every Tuesday, 9 AM–12 PM PT

Foundations

Week 01Tue, Jan 5The landscapeFrom next-token prediction to agents

Introduction to agentic AI

The map before the deep dives. Start with how a model learns its weights and predicts one token at a time, why tokenization sets your inference bill, and how chain of thought became built-in reasoning. See why a hallucination is usually a context problem you can fix, compare pre-training, fine-tuning and in-context learning as ways to give a model your own knowledge, then learn what separates an agent from rule-based automation and preview the planning, tool-use and reflection patterns the rest of the bootcamp builds.

Lecture · Discussion · Live demo

Week 02Tue, Jan 12UnderstandsTokens, embeddings and self-attention

Transformers & attention mechanism

Transformers from the ground up: tokenization, embeddings, positional encoding, and self-attention with keys, queries and values. Hands-on with Sentence Transformers, semantic search, and implementing attention scoring directly.

Lecture · Discussion · Practical exercise

Building blocks

Week 03Tue, Jan 19ChainsLangChain components, tools and LCEL

Mastering LangChain

Core components, output parsers, retrieval and vector stores, chains, tool use, and LCEL composition — the building blocks for assembling a retrieval application that holds together end to end.

Lecture · Discussion · Practical exercise

Week 04Tue, Jan 26RetrievesHybrid search, re-ranking and agentic RAG

Vector databases and agentic RAG

ANN against kNN search, hybrid dense-and-sparse retrieval with metadata filters, MMR re-ranking and HNSW index health — then vector stores as long-term agent memory, with multi-step retrieval loops and semantic caching for cost and latency.

Lecture · Discussion · Practical exercise

Agent design

Week 05Tue, Feb 2Manages contextMemory layers, routing and LangGraph state

Context engineering

Designing what the agent can see: retrieval, memory layers and critique loops; deterministic chains against router agents; LangGraph nodes, edges and state; and the multi-agent architectures — parallel, sequential, hierarchical — that keep a long workflow auditable.

Lecture · Discussion · Practical exercise

Week 06Tue, Feb 9PlansReflection, planning, tools and collaboration

Agentic design patterns

Reflection, planning, tool use and multi-agent collaboration as named patterns: what each one buys, what it costs in autonomy against control, and how to compose them into a hybrid workflow rather than picking one and hoping.

Lecture · Discussion · Practical exercise

Interoperability

Week 07Tue, Feb 16Talks to agentsMCP, A2A and ACP compared

Agentic AI protocols

Why agents need protocols at all — discovery, negotiation and state management across MCP, A2A and ACP, how their scope and message types differ, and which one fits a given workflow. Labs build an MCP client.

Lecture · Discussion · Practical exercise

Week 08Tue, Feb 23ConnectsTools, resources and prompts over MCP

Model Context Protocol

The "USB-C for AI": client-server handshake, resources, tools and prompts over JSON-RPC; exposing context with governance and controlled access; connecting MCP endpoints to planning and multi-agent patterns.

Lecture · Discussion · Practical exercise

Evaluation

Week 09Tue, Mar 2MeasuredBenchmarks, RAGAs and G-Eval

Evaluation of agents

Benchmarks (MMLU, HELM, BBH), text-quality metrics (BLEU, ROUGE, BERTScore), RAG-specific scoring with RAGAs, open-ended evaluation with G-Eval, plus ranking and reasoning metrics — measuring an agent before it meets real users.

Lecture · Discussion · Practical exercise

Capstone

Week 10Tue, Mar 9ShipsA multi-agent application you build and keep

Build a multi-agent LLM application

The capstone. Choose a project track — conversational orchestration, knowledge-enhanced agent, document-aware action agent, or MCP-coordinated collaboration — and ship a production-ready multi-agent application with provided datasets, guides and sandbox templates.

Lecture · Guided lab · Independent project

What you take home

A multi-agent application you build and keep

The program ends in a capstone. Pick a track and ship a production-ready multi-agent application, starting from the datasets, step-by-step guides and code templates in Data Science Dojo's sandbox. A year of sandbox access comes with it, so you can keep building after the last Tuesday.

  • Conversational orchestration
  • Knowledge-enhanced agent
  • Document-aware action agent
  • MCP-coordinated collaboration

Included

1 year
Learner sandbox access
Labs
Agentic AI coding labs
Datasets
Industry-spanning document collections
Certificate
Verified certificate of completion
Tuition & enrollment

Register early and save

Full refund up to five business days before your cohort starts, or transfer to a later date at no charge.

Choose your cohort
WhereOnline · live, Tuesdays 9 AM–12 PM PTIncludesThree hours every Tuesday morning, with the same instructors
30% off · early-bird rate
$2,100$3,000Save $900
30% off ends Nov 24

Then $2,400 (20% off) from Nov 25.

Final rate

This is the last discount tier — enrollment closes when the cohort begins.

Exact dates and early-bird pricing will be announced soon. Register your interest and we'll notify you when enrollment opens.

  • Live sessions with industry experts
  • 1-year access to learner sandboxes
  • Access to the agentic AI coding labs
  • Verified certificate upon completion
  • Access to companion course content for review at your convenience
Reserve your seat
From LLM bootcamp alumni and partners

4.95 from 660+ verified reviews

Read all reviews

Hear it in their own words

  • David MartinsCo-Founder, Notify
  • Dave HortonVP Solutions Engineering, Odaseva
  • Maryam BagheriSr. Data Scientist, Deloitte Business
  • Abrar BhuiyanFull-Stack Software Engineer, ICF
FAQ

Frequently asked questions

How is the agentic AI bootcamp different from the LLM bootcamp?
Mainly the format. The LLM bootcamp runs five consecutive days, in Seattle or live online. The agentic AI bootcamp runs ten weeks, three hours every Tuesday morning, live online. The content overlaps more than the names suggest: both cover RAG, MCP, evaluation and a multi-agent capstone. The LLM bootcamp goes wider on the model layer, with modules on fine-tuning, observability and reference architectures. This one goes deeper on agents, with modules on context engineering, agent design patterns, and agent-to-agent protocols such as A2A and ACP. Both ask only for basic Python. Pick the LLM bootcamp to learn it in one week; pick this one to learn alongside your job.
Who is the agentic AI bootcamp for?
Three groups: engineers who need to take an agent from notebook to production; data scientists and ML engineers who build models and now need the layer around them, from retrieval and orchestration to evaluation; and product leaders and founders who need to scope agent work well enough to tell a week of it from a quarter.
Are there any prerequisites?
Basic Python. That is the bar — you do not need any experience with LLMs, because the early modules cover how language models, embeddings and retrieval work before the course builds agents on top of them. We send preparatory material before the bootcamp starts to help you get ready.
I don't have a background in Python. Can I still join?
Yes. We offer a short introductory course to help you get comfortable with Python.
I am from a non-technical background. Is this bootcamp for me?
It can be, if you are willing to write basic Python — our short introductory course gets you there. Product leaders and founders are one of the three groups this bootcamp is built for: you learn what an agent can automate and how to scope the work. If you are unsure whether it fits, talk to an advisor before you register.
What does the preparatory material include?
A primer on the LLM fundamentals the course builds on:
  • LLM architectures: foundation models, prompts, embeddings, fine-tuning
  • Challenges and risks: prompt brittleness, context limits, security, cost
  • Transformers and attention: self-attention, multi-head attention, tokenization
When will I receive the preparatory material?
About two weeks before the bootcamp starts. You'll receive an email with access details and instructions.
How long is the bootcamp?
Ten weeks, 30 hours in total: a three-hour live session every Tuesday from 9 AM to 12 PM PT.
How much does the agentic AI bootcamp cost?
Tuition is $3,000, and less the earlier you register: $2,100 (30% off) six or more weeks before your cohort starts, $2,400 (20% off) four or more weeks before, and $2,700 (10% off) a week or more before. Inside the last week it is the full $3,000.
Are the sessions live?
Yes. Every Tuesday is a live three-hour session with the same instructors: lecture, discussion and hands-on exercises, with time to ask questions as you go.
What if I'm unable to attend a live session?
You catch up in the companion course. We do not provide recordings of the live sessions, but every key topic they cover is in the companion course as structured lesson clips, so you can review anything you missed at your convenience.
What is included?
  • Live sessions with industry experts
  • 1-year access to learner sandboxes
  • Access to the agentic AI coding labs
  • Access to companion course content for review at your convenience
  • Verified certificate upon completion
Will I receive a certificate?
Yes. Everyone who completes the bootcamp receives a verified certificate of completion from Data Science Dojo.
Do I need my own API keys?
Yes. Bring your own API keys for the model providers used in the labs, such as OpenAI. Your tuition covers the learner sandboxes and the coding labs, but not model usage or cloud subscriptions.
What is the transfer policy for the bootcamp?
You can transfer to a later cohort once at no charge; a second transfer costs a $200 processing fee. Once you have transferred, the registration can no longer be refunded.
What is the refund policy?
Cancel at least five business days before your cohort starts and we refund your registration fee in full. You can also transfer to a later cohort instead. Once you have transferred, the registration can no longer be refunded.
Will I get a job after completing the bootcamp?
There is no job placement. This is upskilling for people already in a role, which is why it runs three hours a week alongside your job. What you do get is a network of fellow attendees, partners and alumni, and once you register we are happy to make introductions to professionals in your field.

Looking to upskill your team?

Corporate cohorts, private training, and volume pricing for teams of five or more.

Talk to us about team training