Design a tamper-evident video chain-of-custody audit system
Company: Axon
Role: Software Engineer
Category: System Design
Difficulty: hard
Interview Round: Technical Screen
Design a scalable, tamper-evident audit logging system for video evidence that preserves an accurate chain of custody. The system must record all actions (upload, view, redact/delete sensitive data, transcribe), never lose an event, and support legal discovery. Specify:
1) functional and non-functional requirements;
2) architecture and storage choices (e.g., append-only log, replication, WORM/immutable storage);
3) event schema, ordering, idempotency, and delivery guarantees (exactly-once vs at-least-once);
4) integrity mechanisms (hash chaining/Merkle trees, digital signatures), encryption and key management, and access controls;
5) scalability, partitioning, and backpressure handling;
6) APIs and query patterns for auditors and investigators;
7) retention policies, legal hold, and compliance considerations;
8) monitoring, audit-of-the-auditor, and disaster recovery;
9) cost/latency trade-offs and capacity planning;
10) testing, verification, and rollout plan.
Overview: Design a tamper-evident video chain-of-custody audit system evaluates requirements, scale assumptions, API/data design, architecture, trade-offs, failure modes, and rollout in a realistic interview setting. A strong answer states assumptions, handles edge cases, explains trade-offs, and shows how to validate the result clearly.
Community answers
Answer by Mohit2289
STEP 1: Clarify Requirements First
At the start of the interview, establish the scope
and constraints of this compliance-critical system:
·
Users & Use Cases: Field Devices/Applications automatically
emitting audit trails (e.g., video started, stream chunk saved); **Internal
Software Services logging system mutations (e.g., AI redaction complete); Legal
Auditors/Investigators** requesting a complete chain-of-custody report for
court discovery.
·
Scale: Millions
of global assets across thousands of agencies. Steady-state **1 to 5 Billion
events/month, with a peak burst capacity of 50,000 events per second
(EPS)**.
·
Latency & Availability: Write path availability must be **99.999%
(Tier-0)**; losing a chain-of-custody event can invalidate evidence in a
legal trial. Event-to-visible query latency can tolerate eventual consistency (**<
5 seconds**). Audit report generation can take up to several seconds/minutes.
·
Correctness & Consistency: Linear timeline correctness per asset is
mandatory. While writes are processed asynchronously for scale, the ledger
state machine must guarantee monotonic event sequencing.
·
Security & Compliance: Strict CJIS, SOC 2 Type II, and FedRAMP High
regulatory compliance. Data must be cryptographically protected against
malicious infrastructure administrators (Insider Threat).
·
Out of Scope (Defensible MVP): We do not store the video binary files here
(handled by a separate WORM Object Storage system); we only ingest, structure,
and cryptographical