CrackML by @ml.with.umang
Interview questions / ML System Design
ML System Design interview question

Design a Second-Pass Feed Ranker

Design a richer second-stage ranker that reranks a few hundred candidates under strict latency.

hardsystem designEvidence 34/1001 source reportTikTok

The 60-second answer

Define what the first stage already guarantees and the second-stage objective, candidate count, feature freshness, and latency budget. Use richer cross features, sequence/session representations, multimodal features, and calibrated multi-task predictions only where they improve ranking enough to justify cost.

Build the answer in this order

1
Frame the problem

Define what the first stage already guarantees and the second-stage objective, candidate count, feature freshness, and latency budget.

2
Design the data path

Use richer cross features, sequence/session representations, multimodal features, and calibrated multi-task predictions only where they improve ranking enough to justify cost.

3
Choose the modeling stack

Apply constrained slate logic for diversity/safety/freshness after or jointly with scoring, and preserve a simpler fallback ranker.

4
Serve, evaluate, iterate

Evaluate incremental NDCG/business lift over stage one, latency/tail latency, calibration, guardrails, and online experiment results.

A useful interview mental model

This is the shape of a strong answer—not a script to memorize.

01Requirements
02Data
03Model / Retrieval
04Serving
05Monitor

Senior-level signal

  • Senior answers discuss distillation/cascades, dynamic early exit, and feature/model cost budgeting per request.
  • Diagnose whether extra complexity adds unique signal or simply relearns first-stage behavior.

What the interviewer is really testing

Product framing, data design, modeling choices, serving constraints, reliability, evaluation, and explicit trade-offs.

Likely follow-up questions

What changes at 10× traffic or data volume?
Which failure mode would you monitor first in production?
How would you evaluate this offline and online before rollout?

Common weak-answer patterns

  • Jumping to a model before defining the product contract.
  • Listing components without bottlenecks, metrics, or failure handling.
  • Ignoring data quality, serving latency, monitoring, and iteration.