How to apply for Research Engineer / Research Scientist, Confidential Computing

About SASH

SASH is a Singapore-based nonprofit focused on building AI safety research capacity and governance expertise across Asia. Its Verification team works on tools to verify agreements about AI, combining technical research with international policy work. The technical team is led by Pascal Berrang, an Associate Professor at the University of Birmingham, and brings experience from the Singapore Government. Policy staff come from Oxford and the Centre for the Governance of AI, with partners including the Future of Life Institute and the University of Oxford.

About the role

This role sits on SASH's Verification team, which prototypes and tests mechanisms to verify what happens inside AI datacenters without requiring model weights to be handed over. The core problem is that the party with something to prove cannot share its weights, which blocks agreements between states, regulators and labs, and companies and their customers. You would build and test confidential computing tools that let these parties prove properties of their systems while protecting sensitive data.

A typical day

A typical day could involve prototyping a verification mechanism using a confidential computing primitive, then testing it against a realistic datacenter scenario. You might also join a call with international research collaborators or policy partners to discuss how a tool would be used in practice. Since the role is remote and the team is small, expect a mix of independent technical work and direct collaboration across time zones.

Who SASH is looking for

• Hands-on experience with confidential computing primitives such as trusted execution environments (TEEs), secure enclaves, or remote attestation
• Ability to prototype verification mechanisms and test them against realistic datacenter or model-training scenarios
• Comfort working at the boundary of technical research and policy, since the team leads international collaborations and engages policymakers
• Interest in AI safety and governance problems, and willingness to work with a small team that mixes technical and policy expertise

Tips for this application

• Read SASH's Verification team output and reference specific verification or confidential computing problems they have written about in your application
• Name concrete confidential computing technologies you have used (for example specific TEE implementations or attestation protocols) rather than speaking about the field generically
• Explain how your work would address the specific constraint in the job description: proving properties of a system without handing over model weights
• Note that the role is remote and the organization is Singapore-based, so state your working hours and any experience collaborating across time zones
• Mention any exposure to AI policy or governance work, since the team works with policymakers and international partners, not only engineers

What to cover in your cover letter

Focus on your specific confidential computing experience and which primitives you have implemented or tested. Explain how you would approach the problem of verifying datacenter behavior without sharing model weights. Show that you understand why this matters for agreements between states, regulators and labs, and companies and customers. Reference SASH's Verification team, its leadership, and its policy partners to show you have read about the organization rather than applying generically.

Research before applying

• Read SASH's published work and any public material from the Verification team, including anything by Pascal Berrang
• Look into the Centre for the Governance of AI, the Future of Life Institute, and the University of Oxford as SASH partners, and understand what they work on
• Study the current state of confidential computing and remote attestation, including known attacks and open problems
• Find out how SASH describes its AI safety research capacity building across Asia, so you understand the broader organizational context

Likely interview topics

Based on the job description, expect questions about:

• How would you design a verification mechanism that proves a property of a model running in a datacenter without revealing the weights?
• Which confidential computing technologies have you worked with, and what are their trust assumptions and limitations?
• How would you test whether a verification prototype actually holds up against an adversarial datacenter operator?
• How do you see verification tools fitting into agreements between states, regulators and labs, or between companies and customers?
• What is your experience collaborating with policy experts or non-technical partners, and how would you handle that in a small team?

Common mistakes to avoid

• Applying with a generic AI safety or machine learning cover letter that does not mention confidential computing or verification
• Claiming expertise in TEEs or attestation without being able to describe specific implementations, trust models, or limitations
• Ignoring the policy side of the role and treating it as a purely engineering position, when the team works directly with policymakers and international partners