LI.FI Analytics Engineer - Take-Home Assessment

Welcome! This assessment consists of three connected tasks that progressively build on each other. You are expected to spend approximately 4–6 hours in total. We value clarity and thoughtfulness over completeness—if you run out of time, document what you would have done differently.

Assume that the company's reporting currency is EUR.

Data Sources

1. LiFi Transfer Events (on-chain, per-block)

   • Source: Dune Analytics (lifi_multichain.LiFiDiamond_v2_evt_LiFiTransferStarted)
   • Fallback provided: To save you time and prevent Dune API rate limits, we have provided a sample of ~20k rows in the /data folder. You can use this directly.
   • Consideration: The sendingAssetId field uses a zero-address (0x000...000) as a placeholder for native tokens (ETH, MATIC, etc.). Think about what this means for downstream joins.

2. Token Prices in USD

   • Source: Dune Analytics (prices.usd) or any free API (CoinGecko, etc.)
   • You must source this data yourself. Automating the extraction via an API is a huge plus, but if you hit rate limits, you can download it manually and place it in the /data folder.

3. External Source [BONUS]

   • Source: E.g., Crypto Fear & Greed Index API (https://api.alternative.me/fng/?limit=0)
   • This is an optional bonus dataset. Focus on the core transfers and prices first.

---

Task 1: Data Model Design

Deliverable: A written description + a simple diagram in the /task_1 folder.

Propose an analytical data model structure suitable for a transformation tool like dbt or SQLMesh.

• Layering: How would you organize your layers?
• Grain & Joins: How do transfers connect to prices?
• Currency: How would you approach EUR conversion?

Task 2: Data Ingestion

Deliverable: Working scripts in the /ingestion folder.

Load the data sources into a local DuckDB database.

• DuckDB can natively read CSVs. You don't need a complex pipeline; a simple Python script or COPY statements work fine. The goal is to get clean, queryable tables.

Task 3: Transformation Project

Deliverable: A working dbt or SQLMesh project in the /dbt_project folder.

Transform the raw data into an analytical model.

• Staging: Clean and type-cast. The bridgeData JSON column needs to be unpacked.
• Marts: Join transfers with prices, calculate volumes (remembering crypto nuances).
• Tests & Docs: Define meaningful tests and descriptions.

---

Submission

Please submit your work as a Git repository (GitHub/GitLab link or a zip archive). If you use AI assistants, that's perfectly fine—just be ready to explain and defend every design decision during the walkthrough session!