Назад
Company hidden
2 дня назад

Staff Engineer, Applied AI

225 000 - 265 000$
Формат работы
hybrid
Тип работы
fulltime
Грейд
senior
Английский
b2
Страна
US
Вакансия из списка Hirify.GlobalВакансия из Hirify Global, списка международных tech-компаний
Для мэтча и отклика нужен Plus

Мэтч & Сопровод

Для мэтча с этой вакансией нужен Plus

Описание вакансии

Текст:
/
TL;DR
Staff Engineer, Applied AI (LLM Agents/MCP): Designing and operating Tinder’s internal agent platform across gateways, connectors, identity, policy, guardrails, observability, and cost with an accent on secure agent actions and reliable production infrastructure. Focus on delegated identity, fine-grained authorization, prompt-injection defenses, agent evaluation, auditability, and incident response for systems handling sensitive data and consequential actions.

Location: Hybrid role requiring in-office collaboration three times per week in Palo Alto, California

Salary: $225,000–$265,000 per year, subject to geographic adjustment if authorized to work outside the listed location.

Company

hirify.global operates a large-scale dating platform focused on human connection, behavioral science, AI, safety, and cultural nuances across more than 190 countries.

What you will do

  • Own the architecture and roadmap for the internal agent platform, including the gateway, MCP connectors, identity, policy, guardrails, observability, and cost controls.
  • Design delegated and agent-owned identities, short-lived scoped credentials, and traceability for actions taken on behalf of users.
  • Build fine-grained access controls, human approval flows, and tool-boundary guardrails against prompt injection, confused-deputy access, data exfiltration, and excessive scopes.
  • Develop traces, audit trails, evaluation systems, regression detection, and per-team cost attribution for agent behavior.
  • Define onboarding, review, versioning, sandboxing, and safe defaults for connectors and third-party MCP servers.
  • Operate the production platform and on-call, managing latency, failure modes, rate limits, quotas, kill switches, incidents, and collaboration with Security and platform teams.

Requirements

  • 8+ years building backend, infrastructure, or platform systems relied upon by other teams, including recent hands-on work with LLM agents, tool use, or MCP.
  • Experience running agent evaluations to improve performance on a specific task.
  • Experience shipping identity or authorization systems, including OAuth 2.0 and OIDC, token exchange or delegation, service identity, and RBAC, ABAC, or relationship-based policy models.
  • Ability to assess agent systems from an attacker’s perspective and explain why prompt-level defenses are insufficient.
  • Experience operating gateways, proxies, or other services in the critical request path, including on-call, observability, and incident response.
  • Strong Python or TypeScript skills and daily use of AI coding tools with a clear understanding of their limitations.
  • Strong communication and collaboration skills, including experience addressing real customer challenges and working with security teams and internal customers.

Nice to have

  • Experience with policy engines or workload identity.
  • Production experience building or operating an LLM gateway, model router, or MCP server.
  • Familiarity with OpenTelemetry and tracing for non-deterministic systems.
  • Contributions to agent identity or MCP authorization standards.
  • Background in application protection, identity and access management, threat detection, or incident response.

Culture & Benefits

  • Hybrid collaboration with required office attendance three times per week.
  • Work closely with internal teams to build practical solutions for day-to-day challenges.
  • Partner with Security, forward-deployed engineers, and platform teams while mentoring engineers contributing to the platform.
  • Own production operations, on-call responsibilities, and incident response for a critical internal platform.

Будьте осторожны

Если работодатель просит войти в их систему, используя iCloud/Google, прислать код/пароль, запустить код/ПО, не делайте этого. Это мошенники. Обязательно жмите «Пожаловаться» или пишите в поддержку. Подробнее в гайде →