Interview Questions › AI Product

AI moderation for comments in regional languages

ShareChat gets millions of comments a day in 15 Indian languages. Design an AI moderation system and explain how you would evaluate it.

Context

Human moderators cannot keep up. Over-removal angers creators; under-removal lets abuse spread. Some languages have little labelled data.

Clarifying questions to ask first

• What policies (hate, harassment, spam)?
• Remove, hide or down-rank?
• How much human review capacity is there?

How to approach it

1. Start from the user problem, not the model: is AI the right tool here?
2. Describe the experience and where the model fits in it
3. Define what "good output" means and how you would evaluate it before and after launch
4. Name the failure modes (wrong answers, bias, privacy, cost) and how you contain them
5. Close with a launch plan and success metrics

Framework: User and problem -> Is AI the right tool? -> Experience design and human-in-the-loop -> Quality bar and evaluation -> Risks, cost and latency -> Staged launch and metrics

Sample answer

• Confidence tiers: auto-action, human review, allow
• Per-language thresholds and evaluation
• Appeals as a feedback signal

1. Labelled eval set per language, including slang and code-mixing
2. Precision/recall by language and policy
3. Appeal overturn rate as a live precision check

Tiered system: auto-hide only at high confidence, human review in the middle band, per-language thresholds tuned on labelled sets. Measure prevalence of violating content seen, with appeal overturn rate and creator complaints as guardrails.

What interviewers look for

• Precision and recall by language
• Time to action
• Appeal overturn rate