Portfolio Projects
The AISL showcase projects — ten builds from reference to-do agent to hospital platform — read as evidence for the portfolio framework.
A portfolio project is an argument that you can ship, and the AISL showcase is a set of worked arguments: ten builds from the community and the team, each with its write-up on the site. Read together with the selection framework, they show what "portfolio-grade" concretely means at three levels of ambition.
Pick the target first
Selecting a Portfolio Project inverts the usual order: choose a domain, find target companies, read their job descriptions and blogs, extract shared problems, and only then pick the project and stack. The AI Engineering Buildcamp carries the same idea into its curriculum — from idea to submission, with cohorts of project ideas (module 1) and a capstone week with evaluation criteria and peer review (module 7).
The reference build
Before building your own, study the bar: To-Do List Agent is the buildcamp's reference project — spec-driven tool creation over an OpenAPI backend, Logfire tracing with per-interaction cost, and 18 pytest tests. It is small on purpose; everything it ships is everything a hiring manager probes.
What the showcase teaches
The member projects cluster into three patterns:
- RAG over real documents: the Cybersecurity Disclosure Agent (raw SEC filings, subsidiary modeling), the Habit Builder Agent (podcast transcription into Qdrant, plus guardrails), and the Intelligent Email Agent (Gmail into Elasticsearch).
- Orchestration: the User Satisfaction Analyst Agent routes between MongoDB and Neo4j specialists, and the guided Multi-Agent Research Assistant builds the orchestrator-searcher-summarizer-fact-checker pattern step by step.
- Production engineering: the AI Data Cleaning Assistant pairs an LLM planner with a deterministic engine all the way to Cloud Run, the Holiday Planning App documents which AI tool did which job, and the Panacea Hospital Management System — no AI at all — shows the RBAC, audit-logging, and soft-delete bar that sensitive-data projects are judged against.
For a first build, the Photo-to-Story AI Pipeline is the deliberately small entry: vision, text, and speech in one week-sized project.
The data behind the method
The field guide supplies the evidence the selection framework runs on: the use-case data says automating manual workflows tops what companies build with (finance, healthcare, legal, and cybersecurity lead the domains), and its seven-step portfolio method turns that data into project candidates with technology chosen last. The record: Field Guide: Portfolio Projects, grounded in the same buildcamp bar as this page.
Where to go next
- Turn the project into applications: AI Engineering Career
- The techniques these projects use: Agents, RAG
- The interview that drills your project: Interview Prep
- The method and its market data: Field Guide: Portfolio Projects
Keep building with a membership
Every topic guide is free to read. Members get structured courses, live workshops, and feedback on real projects.