How do I generate text embeddings?

Embeddings power semantic search and RAG. POST each doc to the embeddings endpoint and build (doc, vector) pairs - in a live cloud workspace. Build it free on HeyDevJob in a live cloud workspace.

AI EngineerJuniorembeddingsopenaivectors

Generate Embeddings for a Document Set

A real ai/ml problem you debug end to end in a live cloud workspace, then show on your portfolio. No tutorial, no toy app - a broken system that behaves like production.

Level

Junior

Time

~20 min

Cost

Free

The scenario

We're building a search-over-docs feature and need vector embeddings for each document. Fill in embed_docs(docs) in /workspace/embed.py so it returns a list of (doc, vector) tuples ready for ingestion into a vector store.

The broken code you start with

embed.py (the unfinished embedder)

def embed_docs(docs):
    # TODO: POST each doc to EMBED_URL, return (doc, vector) tuples
    return []   # nothing is embedded yet

What this teaches you

What you did: Looped through the docs, posted each one to the embeddings endpoint, and assembled (doc, vector) tuples - the shape every vector store (pgvector, Pinecone, Chroma, OpenSearch) takes for ingestion.

Why it matters: Embeddings are the foundation of every RAG pipeline, semantic-search feature, and recommendation system. Knowing the model -> vector shape is the entry ticket to all of it.

In the real world: Production embedders batch (POST a list, not one-at-a-time - 10x throughput), pin the embedding model version (changing models invalidates every stored vector), and store the model name alongside the vector so you can re-embed the corpus when you upgrade.

What you'll practice

Calling an embeddings endpoint per document
Reading the vector out of the response
Building (doc, vector) pairs for retrieval

Why this impresses a hiring manager

This is a real embeddings problem teams hit in production - not a synthetic puzzle.
It shows you can diagnose and fix a AI/ML issue in a live system end to end.
It lands on your portfolio as a scenario a hiring manager can open and click through.

On your portfolio, this becomes

Filled in embed_docs() to call the embeddings endpoint (nomic-embed-text via Ollama, same shape as OpenAI's embeddings API) for each doc and return a list of (doc, vector) pairs ready for a vector store.

Keep going

Restore a Broken LLM API IntegrationAI/ML project AI/ML roadmapStep by step to hired AI/ML interview questionsSTAR answers All AI/ML projectsProjects hub

Build this project free

You're in a real cloud workspace in 30 seconds. Fix it, and it lands on your portfolio.

Start this project →