Bronze to Gold Data Pipeline with Genie Code | Dataford Interview Questions

Generate bronze datasets (transactions, stores, customers) for a retail scenario, introduce data quality issues (such as nulls and casing inconsistencies), transform them into a silver layer, build an aggregate gold view, and construct analysis/EDA notebooks using Genie Code.

Asked in the Vibe Coding Round stage. The interviewer evaluated prompting Databricks Genie Code, inspecting generated code, and transforming records across Medallion architecture layers.

Implement build_retail_layers(transactions, stores, customers). Each argument is a list of dictionaries. Return a dictionary containing cleaned silver records, store-level gold aggregates, and deterministic eda summaries. Normalize identifiers, discard invalid or unreferenced transactions, deduplicate transaction IDs, and report the resulting counts.

Constraints

0 <= len(transactions), len(stores), len(customers) <= 100000
Each input element is a dictionary with the fields used by the function
Identifiers are strings, null, or values convertible to strings
Amounts are numeric or numeric strings; quantities are integer-like
The first occurrence of a normalized transaction ID is retained
Output records and aggregate records must be deterministic

Function Signature

def build_retail_layers(transactions, stores, customers):

Your question is Bronze to Gold Data Pipeline with Genie Code. Start with the requirements in the Question tab.

Run and submit as often as you like. When you're ready, talk me through your approach or go straight to the code.

Problem

Asked in the Vibe Coding Round stage. The interviewer evaluated prompting Databricks Genie Code, inspecting generated code, and transforming records across Medallion architecture layers.

Constraints

0 <= len(transactions), len(stores), len(customers) <= 100000
Each input element is a dictionary with the fields used by the function
Identifiers are strings, null, or values convertible to strings
Amounts are numeric or numeric strings; quantities are integer-like
The first occurrence of a normalized transaction ID is retained
Output records and aggregate records must be deterministic

Function Signature

def build_retail_layers(transactions, stores, customers):

Your question is Bronze to Gold Data Pipeline with Genie Code. Start with the requirements in the Question tab.

Run and submit as often as you like. When you're ready, talk me through your approach or go straight to the code.