Feature Flag Evaluation Engine | Dataford Interview Questions

Given flags.json (flag definitions with key, enabled, default, and rules made of attribute/operator/value conditions), users.json (10,000 users with country, plan, etc.) and expected_decisions.json (answers for 200 users x 7 flags), implement a feature flag evaluation engine: a function to evaluate a single flag for a user and a function to evaluate all flags for a user. Conditions within a rule are ANDed.

Asked in the onsite practical coding stage. Validate against expected_decisions.json first, handle multi-rule conflicts and percentage rollout, then add unit tests.

Use evaluate_flag(flag, user) and evaluate_all_flags(flags, user). A matching rule returns its value; rules are checked in order, and no match returns default. Supported operators are equals, not_equals, in, and not_in. A rule may include percentage, evaluated deterministically from flag.key and user.id.

Constraints

Each flag has a unique string key.
A flag contains enabled, default, and optionally rules.
Conditions within a rule are ANDed.
Rules are evaluated in listed order, and the first matching rule wins.
Supported operators are equals, not_equals, in, and not_in.
Percentage values are integers from 0 through 100.
A percentage rule requires user.id.
The input dataset contains up to 10,000 users and 7 flags.

Function Signature

def evaluate_flag(flag, user):

Your question is Feature Flag Evaluation Engine. 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 onsite practical coding stage. Validate against expected_decisions.json first, handle multi-rule conflicts and percentage rollout, then add unit tests.

Constraints

Each flag has a unique string key.
A flag contains enabled, default, and optionally rules.
Conditions within a rule are ANDed.
Rules are evaluated in listed order, and the first matching rule wins.
Supported operators are equals, not_equals, in, and not_in.
Percentage values are integers from 0 through 100.
A percentage rule requires user.id.
The input dataset contains up to 10,000 users and 7 flags.

Function Signature

def evaluate_flag(flag, user):

Your question is Feature Flag Evaluation Engine. 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.