Trust & Safety Product Operations Specialist

Instawork•San Francisco, CA

19h•Onsite

About The Position

Instawork is hiring a Trust & Safety Product Operations Specialist to investigate safety incidents and work with Product and Legal to prevent future incidents. The role reports to the Director of Trust & Safety and turns findings from cases and data into product feedback and process improvements. It uses AI to query data and build tools, with the specialist responsible for checking the outputs. The specialist handles investigations from intake through resolution, evaluates conflicting accounts and evidence, documents verifiable findings, manages escalated tickets, applies Community Guidelines, and recommends policy revisions where cases reveal gaps. The role also uses natural-language-generated SQL to examine data, identifies fraud patterns and safety trends, tracks and reports Trust & Safety metrics, investigates deviations from targets, and re-reviews samples of closed cases for accuracy and consistency. Product operations work includes testing safety and enforcement features before launch, reporting defects to Product, supporting outreach and adoption tracking, analyzing feedback, and building and maintaining internal tools with AI coding assistants. Tools must be tested before use on live cases and preserve data privacy and human review. The position is based in Instawork’s in-person work environment; the company states it is working together in person five days a week.

Requirements

At least 1 year of experience in Trust & Safety, Operations, or Compliance.
Demonstrated ability to weigh conflicting accounts and reach fair findings under pressure, with a habit of checking personal bias.
Daily professional use of AI to query data and build tools, and working knowledge of its limitations.
Fluency in Excel or Google Sheets and comfort using internal systems to assemble facts.
Experience tracking and reporting operational metrics, with judgment about which metrics may be misleading.
Ownership of each case until the underlying problem is solved.

Nice To Haves

SQL, Python (Pandas), or experience with a BI tool such as Looker or Tableau.
Familiarity with AI coding assistants such as Codex, Claude Code, Cursor, or Replit, or no-code tools such as Zapier.
Experience working with external stakeholders such as clients, legal teams, or law enforcement.
Experience testing or launching product features.
Background in a marketplace or labor platform.

Responsibilities

Run safety investigations from intake through resolution, assess conflicting accounts against evidence, and document findings that another reviewer can verify.
Handle escalated tickets and communicate with professionals, partners, and, when required, legal teams and law enforcement.
Tie case resolutions to the facts and Instawork’s terms.
Apply Community Guidelines consistently and propose revisions when cases expose policy gaps.
Generate SQL from natural language, review query logic, and verify results before acting on them.
Identify fraud patterns and safety trends, and report what the data supports and where it is limited.
Track and report Trust & Safety metrics, including incident rate, time to first action, appeal overturn rate, and repeat-violation rate, against targets set by the Director of Trust & Safety.
Investigate metric deviations and recommend responses.
Re-review a monthly sample of closed cases to assess the accuracy and consistency of findings.
Test safety and enforcement features before launch and provide written defect reports to Product.
Help plan outreach to professionals and partners for new features and track adoption after launch.
Collect and analyze feedback from cases and support channels, and report implications for departmental and company planning.
Build internal tools with AI coding assistants, verify their logic and test outputs before they touch live cases, and maintain data privacy and human review.
Identify manual investigation steps and replace them with automation or tools the specialist builds and maintains.
Own cases until the underlying problem is solved.