Software Engineer III/Senior, AI Gateway

• Own AI traffic policy enforcement: You'll design and implement AI Gateway Traffic Policy Objects-rate limits, usage caps, and access rules specific to AI workloads. These policies exist to prevent runaway costs, misuse, and accidental exposure without breaking legitimate traffic.
• Track AI usage and token consumption: You'll build and maintain systems that accurately measure AI usage-requests, tokens, and related metadata-so customers can understand how their AI systems behave and what they're consuming.
• Make AI behavior observable and explainable: You'll expose clear, trustworthy signals around AI traffic: what was allowed or blocked, which policies applied, and how usage accumulated. When customers ask "what happened?", the gateway should already know.
• Design abstractions that hide complexity: You'll work with product and design to build AI-specific gateway primitives that feel intentional and safe, without leaking provider quirks or infrastructure details into customer workflows.
• Ship systems customers trust in production: You'll collaborate closely with Gateway, Customer Data, and Platform teams to ensure AI usage data, policy enforcement, and billing signals line up-so customers can turn these features on with confidence., This is a remote position for candidates outside of the Bay Area and a hybrid role for candidates within commuting distance to San Francisco. Our Bay Area employees commute to the office on Tuesdays and Wednesdays.

Do you have experience in TypeScript?, * You're comfortable in a statically typed, compiled language such as Go, Rust, C++, or Java (with bonus points for Go)

• You've worked with AI/LLMs and can appreciate their unique brand of edge-cases
• You care about developer experience and thoughtful abstractions
• You enjoy defining system behavior, not just plumbing
• You've thought about retries, limits, and costs before being asked
• You like systems that move complexity from the user to the system

Extra credit if you've worked on:

• AI platforms or inference infrastructure
• API gateways with product-level opinions
• Usage limits, quotas, or billing-adjacent systems
• Customer-facing observability tools, All candidates must be US-based, and legally authorized to work in the United States.

Pulled from the full job description

• Food provided
• Parental leave
• Health insurance
• 401(k) matching
• Vision insurance
• Dental insurance, * Health stuff that actually matters. Full premiums covered on base healthcare, dental, and vision for you. Half covered for your dependents. Mental health and well-being support included, because taking care of your brain is as important as taking care of your teeth.
• Retirement matching that doesn't suck. 401(k) with 100% match up to 3% of your salary and 50% match up to another 2%. Future you will appreciate present you.
• Actually flexible time off. We say "open, flexible vacation policy" and actually mean it. Take the time you need. Your manager will bug you if you're not taking enough.
• Parental leave that's realistic. Up to 16 weeks if you give birth, up to 8 weeks for new parents (birth, adoption, fostering-however your family grows).
• Money to keep growing. Annual professional development budget for books, courses, conferences, or whatever helps you level up. Plus an annual home office/desk stipend to make your workspace not terrible.
• Work from wherever. Co-working space stipend if you want to get out of your house but aren't near our SF office.
• Lunch on us. 2x+ per week for employees onsite at our San Francisco office. Free food tastes better.
• Company offsites. Twice a year we get the whole team together. It's part strategy, part bonding, part excuse to hang out with Bufo (the toad).
• Regular feedback and fair compensation. Bi-annual reviews to make sure you're getting real feedback and staying competitively compensated. No surprises, no waiting around for performance conversations.

ngrok is an all-in-one cloud networking platform that secures, transforms, and routes traffic to services running anywhere. Instead of cobbling together nginx, NLBs, VPNs, model routers, and oodles of other tools, developers solve every networking problem with one gateway. Doesn't matter if they're sharing localhost or running AI workloads in production. We're trusted by more than 9 million developers at companies like GitHub, Okta, HashiCorp, and Twilio. What started as a way to put your local app on a public URL has grown into a universal gateway for API delivery, AI inference, device fleets, and site-to-site connectivity. It's the same ngrok that millions of developers have loved and leaned on every day for years, now with the power to run production traffic at scale. A few things you should know: * We are obsessed with our pets, Viper sunglasses and Bufo (yes, the toad) * We have a designated Chief Emoji Officer - they are vital to our success! * We like software that's serious and culture that's not Still reading? Good. There's more below worth your time. About the AI Gateway Team Our AI Gateway team builds the systems that define how AI traffic is identified, controlled, and understood as it passes through ngrok. We own the AI-specific control plane at the gateway layer: policies, usage tracking, and enforcement that sit directly on live customer traffic. Our systems must behave correctly under real-world conditions-traffic spikes, unexpected model behavior, misconfigured policies, and customers asking, "Why was this blocked?" or "Where did my tokens go?"