TryCrucible — Prove your AI engineering skills

Scenario

A law firm has 150 documents across three formats: PDF extracted text, markdown memos, and HTML case notes. Build a retrieval system that normalises all formats and answers 15 natural language questions accurately with source citations.

Expected output

Accurate answers to 15 test questions with correct source document IDs across all three formats

Dataset

📦

Dataset files

Scoring rubric

Clarity5%

Ingestion pipeline is documented and reproducible

LLM usage20%

Prompt handles multi-source context cleanly without exceeding context limits

Robustness10%

Handles malformed documents, encoding issues, and missing metadata

Correctness25%

Answers match ground truth for all 15 questions, source IDs are correct

Architecture20%

Format-agnostic ingestion pipeline, clean normalisation before embedding

Language-free evaluation

Build your solution in any language or framework — Python, TypeScript, Go, Rust, Java, C#, or anything else. The dataset artifacts may be in one language; your implementation does not need to match. TryCrucible evaluates the behaviour of your system, the quality of your AI workflow, your verification strategy, and the reproducibility of your submission — not your language choice.

Submission requirements

A public GitHub repository link
A Dockerfile in the repo root — any language or framework; the evaluator builds and runs your container
Your solution reads test_inputs.json from the working directory and writes results.json — standard I/O contract across all challenges
A decisions.md — 3–5 sentences on the key architectural and AI-workflow choices you made
The system must be fully reproducible — we clone, build, and run it against real test inputs

Evaluation contract

When you submit, the evaluator runs these steps in order:

1Clone your public GitHub repository
2Build your container from the Dockerfile in the repo root
3Mount test_inputs.json into the working directory
4Run your solution in a network-isolated sandbox (5 min limit, 512 MB RAM)
5Read results.json from the working directory
6Score correctness against hidden ground truth, then score architecture, AI workflow, robustness, and clarity

// I/O contract — same schema for every challenge

Input (provided by evaluator)

// test_inputs.json
[
  { "id": "t1", "input": { ... } },
  { "id": "t2", "input": { ... } }
]

Output (written by your solution)

Create a free account to start

Already have one? Sign in

Build a RAG pipeline over multi-format documents