- Prototyping to reach signs of life on an initial idea
- Navigating the design space of system architecture and other hyperparameters
- Designing and implementing efficient and reliable systems (training and inference pipelines)
- For more senior candidates: direction setting on code code architecture, planning and leading projects
- Reliable and trustworthy results: meticulous, good experimental design, epistemic self-awareness and transparency
- Prioritization: be scrappy while balancing appropriately with meticulousness
- Keep low-level planning connected to high-level goals
- Generative: come up with original, productive ideas for unblocking progress
- Deep domain knowledge in some area of ML research, such as interpretability, RL, basic systems performance / GPUs
- Strong programming ability, including navigating trade-offs between prototyping speed and maintainability
- Strong communication skills, low ego, openness to giving and receiving feedback
- High self-bar for final results
- Ability to learn quickly from teammates and mentors
Skills Required
- Deep domain knowledge in ML research
- Strong programming ability
- Strong communication skills
What We Do
Transluce is an independent research lab that builds open, scalable technology for understanding AI systems and steering them in the public interest. Transluce means to shine light through something to reveal its structure. Today’s complex AI systems are difficult to understand—not even experts can reliably predict their behavior once deployed. Given AI's extraordinary consequences on society, we need scalable and open analyses of the capabilities and risks of AI systems. We are building open source, AI-driven tools to understand and analyze AI systems. We will apply these tools to open-weight models, so the world can vet our analyses and improve their reliability. Once our technology has been vetted, we will work with frontier AI labs and governments to ensure that internal assessments reach the same standards as our publicly vetted procedures. Email: [email protected]
