← All courses

AI & Data

Build with LLMs (OpenAI/Anthropic API)

Use AI like a senior engineer, not a tourist.

Duration
8 weeks
Format
2 live sessions + 1 lab per week
Track type
Cohort
Open seats
7 across 1 cohort
Apply for this course →See cohort calendar

About this course

Why students take this

For engineers building real products with LLMs. Eight weeks of API-driven AI development with OpenAI, Anthropic, and open-source models.

We cover prompt design, structured outputs, function-calling, retrieval-augmented generation (RAG), evaluation, and production deployment. By the end you ship a real AI feature integrated into a real app.

Production AI engineering

  • Provider APIs: OpenAI, Anthropic Claude, Gemini
  • Structured outputs + tool / function calling
  • RAG: chunking, embeddings, vector databases (Pinecone, pgvector)
  • Evals: how to test LLM output without crying
  • Deployment, observability, cost control, fallbacks

Outcomes

What you’ll walk away with

Every outcome below is something a graduate can demonstrate on the last day of class — not a vague promise.

  • 01

    Build production-ready AI-powered features

  • 02

    Write prompts, evals, and guardrails like a senior

  • 03

    Ship a portfolio AI project that demonstrates real skill

Syllabus

How the 8 weeks actually unfold

We move in deliberate phases. Each one builds on the last, so by week 8you’re working on real briefs — not still typing your first hello-world.

  1. W1

    LLM mental model

    Tokens, context windows, sampling.

    You can…
    • Explain how LLMs actually work
    • Reason about cost + latency
  2. W2

    API basics — OpenAI, Anthropic, Gemini

    Streaming, retries, fallbacks.

    You can…
    • Call provider APIs cleanly
    • Handle streaming + retries
  3. W3

    Structured outputs + tool use

    Function calling, JSON schemas.

    You can…
    • Get reliable JSON output
    • Wire LLMs to your tools
  4. W4

    RAG basics

    Chunking, embeddings, vector search.

    You can…
    • Build a basic RAG pipeline
    • Pick chunking + embedding strategies
  5. W5

    Vector databases

    pgvector, Pinecone, Qdrant.

    You can…
    • Choose a vector store for your use case
    • Manage embedding lifecycle
  6. W6

    Evals + observability

    Testing LLM output.

    You can…
    • Write evals for an LLM feature
    • Detect regressions in production
  7. W7

    Cost + latency control

    Caching, model routing, batch.

    You can…
    • Cut LLM cost by 50%+
    • Keep p95 latency under control
  8. W8

    Production deploy

    Auth, rate limits, guardrails.

    You can…
    • Ship a production AI feature
    • Protect against prompt injection

Who this is for

A great fit if you’re…

  • Engineers wanting to add AI to their stack
  • Founders building AI-first products
  • Analysts moving from spreadsheets to Python

Tools you’ll use

The stack you’ll get fluent in

  • OpenAI, Anthropic, Google APIs
  • Python, Jupyter, Pandas, scikit-learn
  • Vector databases, RAG pipelines

Format: 2 live sessions + 1 lab per week. Heavy hands-on coding.

FAQ

Questions students ask before joining

Do I need to know Python?+

Helpful but not required for the intro courses. The Data and LLM courses assume Python basics.

Do I need a GPU?+

No — we use cloud APIs. Your laptop can be modest.

Will I use my own data?+

You can — the LLM and Data courses both encourage bringing your own dataset for the final project.

Are these tools free to use?+

Most have free tiers. We provide API credits during cohort for paid features.

Also in AI & Data

Other courses in this track

AI for Everyone — ChatGPT, Claude, Gemini

2 weeks

Course details →

Prompt Engineering for SMEs

2 weeks

Course details →

Python for Data Analysis

12 weeks

Course details →

Next step

4-minute application. Quote in 24 hours.

Tell us about your background and goals. We’ll send the syllabus, fee plan, and a cohort that fits.

Apply for Build with LLMs (OpenAI/Anthropic API) →