Software Engineer, Trust & Safety - OpenRouter

About OpenRouter

OpenRouter is the leading AI routing and infrastructure layer that enterprises use to access, manage, and optimize the best large language models across providers—without lock-in, capacity constraints, or unnecessary cost. We power the most advanced AI teams in the world by giving them the flexibility to move fast, scale confidently, and stay future-proof as models evolve.

As enterprise adoption of AI accelerates, OpenRouter sits at the center of how organizations operationalize LLMs across research, product, and production workloads.

About the Role

We’re hiring a Software Engineer on our Trust & Safety team to own the detection, enforcement, and review systems that keep abuse, fraud, malicious users, and illegal content off the platform. Millions of LLM requests and a large volume of payments flow through OpenRouter every day, and maintaining the safety, integrity, and trust of our platform is a priority. You’ll build the systems that detect abuse and fraud, the heuristics for automated action and abuse mitigation, the tooling our reviewers act through, and the guardrails that keep us from acting on the wrong account.

You will be responsible for proactive abuse-prevention architecture for a new layer of the AI stack while the abuse patterns are still being invented. The threat model changes quickly, and the systems you build have to catch novel abuse without adding friction for legitimate developers and enterprises. Done well, this work protects millions of daily requests, our provider relationships, and the trust that enterprises place in OpenRouter.

This is a hands-on IC role with broad surface area spanning the inference path, the payments path, internal review tooling, and the data platform underneath all of it. There is little process between you and shipping, and your work has direct consequences for real users, so the bar for judgment is high.

What You’ll Do

• Build and operate technical systems across signup, payments, and usage to detect abuse and fraud early.
• Own the technical enforcement pipeline end to end, from detection proposing a candidate to a human reviewing it to restrictions landing across our systems.
• Build internal tools that support investigation and enforcement decision-making (case queues, evidence summaries, bulk review and enactment, and alerting mechanisms).
• Ship technical solutions for content safety on the inference path: illegal content detection and reporting, KYC systems, or external intelligence sources to detect and stop fraud and abuse.
• Make abuse uneconomical by building systems, tools, and heuristics for quick detection and scalable enforcement.
• Investigate incidents directly in the data and build the analytics plane to establish patterns and separate real abuse from false-positive clusters.
• Build defensible monitoring for risk, abuse spikes, and fraud, while reducing false positives.
• Set the technical direction for abuse prevention as the platform grows and define patterns for other engineers.
• Work with data scientists on feature exploration and training of risk and abuse ML models.

What We’re Looking For

• 4+ years building and operating production systems, ideally including work in trust & safety, fraud, payments risk, security, or anti-abuse.
• Proficient in React, TypeScript, Next.js, and JS runtimes.
• Ability to write SQL against large event datasets and reason about base rates, precision, recall, and the cost of wrong decisions.
• Sound judgment when working with incomplete evidence.
• High agency and a bias toward action; spotting problems, iterating, and shipping without waiting for tickets.
• AI-forward workflow: using coding agents often, finding ways to automate work, and holding informed opinions on what works.
• Comfortable in a small, fast-moving environment where team boundaries are fluid.
• Discretion and resilience when reviewing sensitive user data or discussing disturbing content.
• Strong written and verbal communication skills to explain incident findings, detection rules, and systems.
• Motivated by adversarial problems where the counterparty adapts continuously.

Nice to Haves

• Experience with payments fraud tooling (Stripe Radar, chargeback/dispute flows, crypto payment risk), or identity and KYC systems.
• Experience with LLM-specific abuse: jailbreaks, prompt injection, key theft and resale, shared or scraped credentials, or automated account farming.
• Experience operating large-scale analytical datastores (e.g., ClickHouse, BigQuery) and observability platforms.
• Familiarity with compliance and reporting obligations for hosted AI, including illegal content reporting and provider policy requirements.
• Existing user of OpenRouter or creator of active side projects in AI products, infrastructure, or developer tooling.