Tinfoil (YC P25) is hiring an applied AI security engineer in San Francisco
Tinfoil: https://tinfoil.sh/ is a security research lab building an AI cloud for private, verifiable computation. Our customers include DuckDuckGo, Mozilla, financial institutions, AI labs and startups. Frontier labs, Google Deepmind, Trail of Bits, Redhat and Canonical, have also used us for research.
As attacks on infrastructure grow more powerful, our goal is to bring state-of-the-art defenses to individuals and organizations as quickly as possible, without making them trade away performance or ease of use. Our mission[1] is to create a private garden for thought, where people keep control of their own agency, and we keep evolving our research[2] to stay ahead of rapidly changing threat models.
About this role
As an Applied AI Security Engineer, you’ll help companies deploy AI on Tinfoil when security or privacy is a hard requirement. Our customers include enterprises in regulated industries, AI labs and startups that need secure compute, and privacy-first product companies. You’ll partner with Sales to guide strategic customers from technical discovery through deployment. Then you’ll bring what you learn back to Product, Engineering, and Research so that what we build next reflects what customers need. The role is equal parts product engineer, security advisor, field expert, and connective tissue.
This role is based in our San Francisco office, in person at least 3 days a week.
With customers, you’ll become a trusted technical advisor. You’ll turn their threat models and privacy requirements into a working architecture, and show them how to verify Tinfoil’s guarantees themselves. You will help strategic customers understand the value of Tinfoil and paint the vision on how they can successfully integrate and deploy Tinfoil into their technology stack. When they hit problems, you’ll get hands-on: reproducing their environment, digging into the issue, and building working solutions with them.
With product, you’ll get involved early to build deep expertise on new capabilities as they develop. You’ll help test new products to understand their strengths and limitations, and build demos for public or customer-facing use. And throughout, you'll help shape our roadmap by synthesizing what you're learning across customers and products into clear signal about what’s working, what isn’t, and what matters most.
To succeed, you’ll need the technical depth to push the limits of what Tinfoil can do and a strong understanding of our product and security model. You’ll also need to explain complex capabilities clearly to many different audiences in a world where threat models keep changing.
Responsibilities:
1. Serve as the primary technical advisor to strategic customers from discovery through production deployment, working across Sales, Product, Engineering, and Research to make them successful
2. Turn each customer’s threat model and security or privacy requirements into a deployment architecture built on Tinfoil
3. Build pilots, prototypes, and eval suites that prove Tinfoil meets a customer’s performance and security bar
4. Lead architecture and security reviews with the customer’s engineering, security, and compliance teams, including showing them how to verify Tinfoil’s guarantees themselves
5. Get hands-on when customers hit new problems: reproduce their environment, find a solution, and turn it into an approach other customers can reuse
6. Create reference architectures, playbooks, and public content (blogs, talks) and talks at conferences on deploying AI with high-assurance security
7. Find patterns across customers (adoption blockers, capability gaps, compliance needs) and work with Product and Research to get them on the roadmap
8. Travel occasionally to customer sites for workshops and implementation support
You may be a good fit if you have:
1. 2+ years of experience in technical roles such as software engineer, forward deployed engineer, security engineer or technical product manager, with a desire to work closely with customers. Former technical founders are also encouraged to apply.
2. Strong programming skills in Python or TypeScript and experience building production applications.
3. A solid background in security, networking, or infrastructure, and comfort explaining threat models and security guarantees to skeptical audiences.
4. Hands-on experience building with LLMs, such as agents, workflows, or LLM-powered products, including designing evals and prototypes.
5. Strong communication skills: you can explain complex technical and security concepts clearly to founders, engineers, security and compliance teams, and executives, with a low ego and a collaborative approach.
6. A track record of going deep on technical problems and turning what you learned into materials, tools, or training that helped others succeed.
7. The ability to quickly build expertise in new domains and work through ambiguity, finding simple solutions to complex problems
8. Passion for building AI systems with privacy, verifiability, and control as core properties
Bonus:
1. Experience with applied cryptography or hardware security
2. Experience deploying software in regulated industries (finance, cybersecurity, government, biotech) or guiding customers through security reviews
3. Experience with model serving or inference infrastructure
4. Public technical writing or speaking
We encourage you to apply even if you don’t meet every single qualification!
If this sounds like a good fit, please reach out to work@tinfoil.sh or apply here: https://tinfoil.sh/join.