Label and structure real production network data — logs, configs, telemetry, packet captures — so ML models can learn how networks fail and recover. The client wants working Tier 3 escalation engineers, not architects or SOC analysts, at 30–40 hours a week on Pacific time.

Summary, eligibility guidance and screening preparation are prepared by Sidequest from the platform's posting. Check the source listing before applying.

What the work actually involves

You spend your day inside data pulled from deployed enterprise networks: syslog streams, controller event logs, device configs, SNMP and streaming telemetry, and packet captures. Your job is to say what happened and why — classifying behaviors and anomalies, tagging incident patterns, marking where a root cause is genuinely visible in the data versus where it is only inferable. Because the labels feed model training, disagreements about taxonomy matter: you will be asked to help define schemas, argue for or against a category, and flag cases where the existing label set forces a wrong answer.

This is annotation work performed at an engineer's level of judgment. The distinction the client draws matters — they want people who have been the final escalation on a broken retail WLAN at 2am, not people who drew the reference architecture. Wireless depth is treated as a hard requirement rather than a nice-to-have: Cisco WLC, Aruba, or Meraki controllers, 802.1X/RADIUS failure modes, roaming and RF troubleshooting.

What the platform screens for

Mercor's screen is AI-led and follows up on specifics. Expect it to push past tool names into failure modes: what a particular log line meant, how you distinguished a client-side supplicant problem from a RADIUS timeout, what the capture showed that the dashboard did not. Vertical experience with end-customer enterprise networks — retail, healthcare, finance, manufacturing, campus, hospitality — is probed directly, as is whether your current role is genuinely maintainer work. Candidates whose recent experience is design, cloud/SRE, SOC, or helpdesk are commonly screened out at this stage.

Logistics

Remote, 30–40 hours per week, with substantial overlap with the Pacific business day
Direct interface with the client team, so live calls are part of the arrangement — not purely async
Individual 1099 contract paid to a personal account; no corp-to-corp, no agency billing
Must be authorized to work in the US or Canada without sponsorship
Observed rate for this listing is $50–70/hr; bands on Mercor vary by assessment outcome and are not guaranteed

Network Engineer - Data for Autonomous Systems annotation

What the work actually involves

What the platform screens for

Logistics

Interview prep — 10 questions