Startups.com·Machine Learning Engineer·Onsite - System Design / Architecture·Senior

SeniorPrefer not to say

Jun 2026Remote

Summary

System design round at Startups.com for an ML Engineer role, focused entirely on building an AI-powered database performance advisor. One long, dense question that kept branching into sub-problems every time I thought I'd covered it.

Questions Asked (1)

Design a system that accepts natural-language questions about database performance, translates them to SQL, executes the queries safely against the database or its metadata, analyzes the results, and surfaces optimization recommendations to the user. Walk through the NL-to-SQL layer, execution controls, result analysis, recommendation generation, evaluation, and observability.

System DesignTechnical Trade-offsRoot Cause Analysis

Author's notes

This felt like four system design questions stapled together.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Structure your answer around the end-to-end pipeline: NL-to-SQL translation, safe execution, result analysis, and recommendation generation. Emphasize safety and evaluation throughout, and discuss trade-offs between accuracy, latency, and cost. Conclude with observability and monitoring to ensure reliability and continuous improvement.

Pro tip: Highlight the importance of a human-in-the-loop for high-risk queries and the use of synthetic data for evaluation, showing you understand production ML challenges.

1. NL-to-SQL Translation

Choose a model (e.g., fine-tuned LLM) to convert natural language to SQL, using schema-aware prompting and few-shot examples. Discuss handling ambiguity and incorporating user feedback.

2. Safe Execution

Implement query validation, read-only permissions, and resource limits to prevent harmful queries. Use a query planner to estimate cost and reject expensive queries.

3. Result Analysis

Analyze query results and execution plans to identify performance bottlenecks (e.g., missing indexes, full scans). Use heuristics and ML models to detect anomalies.

4. Recommendation Generation

Generate actionable optimization suggestions (e.g., add index, rewrite query) with explanations and confidence scores. Prioritize recommendations by impact.

5. Evaluation & Observability

Define metrics (e.g., translation accuracy, recommendation acceptance rate) and set up logging, tracing, and dashboards. Use A/B testing and user feedback to improve.

Key Points to Mention

Schema-aware NL-to-SQL with few-shot prompting and fine-tuning
Query validation, read-only access, and resource limits for safety
Execution plan analysis and anomaly detection for performance insights
Actionable recommendations with confidence scores and explanations
Evaluation metrics: translation accuracy, execution success, recommendation acceptance
Observability: logging, tracing, monitoring, and feedback loops

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.