Databricks·Software Engineer·Onsite - System Design / Architecture·Senior

Senior

Jun 2026

Summary

Databricks system design round, one big question about building a brokered bookstore service with no inventory of its own. Pretty meaty problem that touched a lot of ground in one shot.

Questions Asked (1)

Design a bookstore service that holds no inventory itself. Customers request books, the service fans out to multiple vendors who respond with bids (price, availability, delivery ETA), and then the service ranks and presents the best offers. Cover vendor onboarding and catalog normalization, request fan-out with timeouts and partial results, bid scoring, order placement and settlement, and reliability under slow or failing vendors at high QPS.

System DesignAPI & IntegrationsTechnical Trade-offs

Author's notes

This one sprawled in every direction and I underestimated how much ground there was to cover.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by clarifying requirements and scale (QPS, latency SLOs, vendor count), then sketch a high-level architecture with separate services for vendor onboarding, request fan-out, bid scoring, and order settlement. Dive into the fan-out mechanism with timeouts and partial results, and discuss trade-offs in bid scoring and reliability patterns like circuit breakers and bulkheads. Conclude with data consistency and failure handling for orders and settlements.

Pro tip: Emphasize idempotency and exactly-once semantics for order placement and settlement, as duplicate orders or payments are catastrophic in a multi-vendor marketplace. Also, discuss how you'd handle vendor bid manipulation (e.g., fake low bids) through reputation scoring and anomaly detection.

1. Clarify Requirements and Scale

Ask about expected QPS, number of vendors, latency SLOs, and consistency requirements. Define functional and non-functional requirements to scope the design.

2. High-Level Architecture

Outline core components: API gateway, vendor onboarding service, catalog normalization service, request orchestrator, bid scorer, order service, and settlement service. Describe data stores and communication patterns.

3. Vendor Onboarding and Catalog Normalization

Explain how vendors register, provide catalogs, and how you normalize book metadata (ISBN, title, author) across vendors. Discuss schema mapping, validation, and periodic syncs.

4. Request Fan-Out and Bid Scoring

Detail the fan-out mechanism: how to query multiple vendors concurrently with timeouts, handle partial results, and aggregate bids. Describe bid scoring criteria (price, ETA, vendor reliability) and ranking algorithm.

5. Order Placement, Settlement, and Reliability

Cover order flow: selecting best bid, placing order with vendor, handling failures and retries with idempotency. Discuss settlement (payment to vendor) and reliability patterns like circuit breakers, bulkheads, and rate limiting to handle slow/failing vendors at high QPS.

Key Points to Mention

Use asynchronous fan-out with timeouts and partial results (e.g., via message queues or reactive streams) to avoid blocking on slow vendors.
Normalize vendor catalogs using a canonical book schema (e.g., ISBN-13) and handle conflicts with versioning or vendor-specific mappings.
Score bids using a weighted formula (price, delivery ETA, vendor rating) and allow configurable weights; consider real-time adjustments based on vendor performance.
Ensure idempotent order placement and settlement using unique order IDs and idempotency keys to prevent duplicate charges.
Implement circuit breakers, bulkheads, and rate limiting per vendor to isolate failures and maintain high QPS under vendor degradation.
Use eventual consistency for catalog updates and strong consistency for orders/settlements; discuss trade-offs and compensating transactions.

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