Footprint
Member of Technical Staff (Security)
$180K–$260K
New York City +1 · On-site · Staff · Full-time · Posted Aug 25, 2026
- Location
- New York City
San Francisco - Experience
- 5+ years of hands-on security engineering.
About the job
Own and harden Footprint’s product security across its identity vault, AI-agent sandbox, end-user data flows, document pipeline, and security controls for compliance audits.
What you'll do
- Assess and secure the Nitro Enclave-backed vault, including encryption, key management, audit-ledger integrity, and cross-tenant isolation.
- Set and enforce security standards for Percy’s isolated Modal sandboxes, live-PII access, agent audit logs, and end-user data paths.
- Run technical penetration tests, provide audit evidence and remediations, and drive vulnerability fixes and security tooling.
What they ask for
- 5+ years of hands-on security engineering in application security, infrastructure security, or both.
- Strong understanding of encryption, key management, and access control for production systems holding PII.
- Knowledge of code-execution escape risks, least privilege for ephemeral compute, and cross-tenant leaks; offensive security experience such as penetration testing, threat modeling, or vulnerability research.
- Hands-on AWS security experience with IAM, VPC, KMS, CloudTrail, and GuardDuty; experience with a SOC 2 or PCI DSS audit.
Nice to have
- Experience securing AI/LLM agents, autonomous code execution, or trusted execution environments such as AWS Nitro Enclaves or Intel SGX.
- Experience with Rust, security tooling, document fraud detection, adversarial ML robustness, or KYC and identity verification.
Pay
$180K – $260K (base salary)
Tools
Rust, TypeScript, Postgres, AWS (Nitro Enclaves, KMS, IAM, VPC, CloudTrail, GuardDuty), Modal, Vercel
AI focus
Secure Percy’s AI-powered agents running in isolated Modal sandboxes against live PII, including sandbox isolation, token scoping, network egress, and agent audit-log integrity.
Model-written summary · the original posting is authoritative