Infer user intent from typing in real time
Company: Microsoft
Role: Machine Learning Engineer
Category: ML System Design
Difficulty: medium
Interview Round: Onsite
## Scenario
You’re building an AI feature that observes a user’s **typing stream** in an editor/search box and predicts the user’s **intent** in real time.
This is **not** classic autocomplete of the next few characters; instead, you need to infer higher-level intent such as:
- “writing an email reply” vs “searching inbox” vs “creating a meeting agenda”
- “asking for code snippet” vs “asking for explanation”
- “trying to insert a table” vs “formatting a heading”
## Task
Design an ML system that infers intent from typing + context.
### Requirements
- Streaming / low latency (updates as the user types).
- High privacy requirements (typed text can be sensitive).
- Must degrade gracefully when uncertain.
- Should support personalization.
### What to cover
- Input signals and features (text + UI context).
- Model architecture and serving (on-device vs server).
- Labeling strategy.
- Metrics and evaluation.
- Failure modes and safety.
Quick Answer: This question evaluates machine learning system design competency in streaming, low-latency intent inference from typing and UI context, covering privacy-preserving modeling, personalization, labeling strategy, metrics, and safety/failure modes.