Databricks AI Bootcamp Capstone Ideas

Key takeaways#

• Databricks AI Bootcamp Capstone is a collection of project briefs for building data apps with Spark, Databricks Apps, retrieval, and AI agents.
• The repository requires each project to include a Spark data pipeline, at least one third-party API, unstructured data processing, a frontend, and an agent that can search, retrieve, and take actions.
• The ideas cover movie planning, trip planning, research copilots, stock analysis, and other applied AI workflows.
• It is a useful resource for students, bootcamp instructors, and builders who want concrete agent-app specs rather than vague project prompts.

What it is#

Databricks AI Bootcamp Capstone is an educational repository of capstone project ideas. Instead of presenting one finished application, it gives learners several structured app briefs that combine data engineering, retrieval, agent tooling, and a Databricks App frontend. The requirements push each project beyond a chatbot demo: students need a real Spark pipeline, a third-party API, unstructured data, a frontend, and an AI agent that can use tools and write back to data.

Project pattern#

Each capstone idea includes suggested APIs, core data tables, unstructured content to embed, retrieval behavior, and actions the agent should be able to take. The movie night planner uses TMDB data, group ratings, plot summaries, reviews, watchlists, and recommendations. The trip planner uses geocoding, weather, air quality, Wikimedia data, itineraries, activities, and packing lists. The research copilot uses papers, authors, collections, reading progress, notes, citations, and learning goals. The stock market assistant uses companies, tickers, price snapshots, news, notes, and analysis reports.

Why this is useful#

Many AI app tutorials stop at retrieval-augmented generation. These capstone specs ask learners to build the parts that make a product feel real: data models, ingestion, embeddings, semantic retrieval, user state, external API calls, frontend flows, and agent actions. That makes the repo useful even outside a Databricks class. A builder can pick one idea and treat it as a product requirements document for an internal hackathon or portfolio project.

Fit for OpenTools users#

The strongest audience is builders learning how to turn AI agents into full data applications. Databricks users get the most direct value because the requirements assume Spark and Databricks Apps. Still, the project structures transfer to other stacks. The movie, trip, research, and finance patterns can be implemented with any warehouse, vector store, web framework, and model provider if the learner keeps the same core constraints.

Limitations#

The repository is not a polished course with videos, grading, or completed reference implementations. It is a set of capstone ideas and requirements. Learners still need to choose APIs, design schemas, write pipelines, build the UI, and implement agent tools. That is also the point: the resource is valuable because it defines practical projects that require real engineering decisions.