Senior Research Engineer

FAR.AI•Berkeley, CA

5d•$225,000 - $350,000

About The Position

FAR.AI is hiring an experienced machine learning engineer or software engineer interested in transitioning to AI safety research. As a Member of Technical Staff, the Senior Research Engineer will join an existing research workstream and lead projects focused on detecting and preventing deception, preventing catastrophic misuse, or accelerating research through improved frameworks and infrastructure. Additional workstreams may become available as FAR.AI expands its research portfolio. Candidates are expected to have significant software engineering experience, fluency in Python, a results-oriented approach motivated by impactful research, and experience mentoring engineers or scientists in engineering skills. They should also bring expertise in at least one of three areas: machine learning, high-performance computing, or technical leadership. The machine learning focus involves large-scale transformer training and requires substantial experience with frameworks such as PyTorch or JAX, along with knowledge of linear algebra, calculus, probability, and statistics. The high-performance computing focus involves cluster orchestration, distributed systems, and code optimization and profiling. The technical leadership focus involves designing large-scale software systems and managing small teams and project roadmaps. The position is full-time at 40 hours per week and can be remote or in person in Berkeley, California, or Singapore. FAR.AI can hire remotely in most countries and sponsors visas for in-person employees. Employees based in the USA or Singapore are employed by FAR.AI; those elsewhere are employed through an employer-of-record organization or as contractors. Compensation is $225,000–$350,000 per year, adjusted based on salary bands, experience, and location. Work-related travel and equipment expenses are covered, and catered lunch and dinner are provided at the Berkeley office.

Requirements

Significant software engineering experience, evidenced by prior work experience and/or open-source contributions.
Fluency working in Python.
Results-oriented and motivated by impactful research.
Prior experience mentoring other engineers or scientists in engineering skills.
Expertise in at least one of the following areas: machine learning, high-performance computing, or technical leadership.
For the machine learning area: substantial experience training transformers with common ML frameworks such as PyTorch or JAX, and good knowledge of basic linear algebra, calculus, vector probability, and statistics.
For the high-performance computing area: power-user experience with cluster orchestrators such as Kubernetes or SLURM; experience building high-performance distributed systems, such as multi-node training or large-scale numerical computation; and experience optimizing and profiling code, ideally including GPU code such as CUDA kernels.
For the technical leadership area: experience designing large-scale software systems, such as architecting greenfield software or leading a major refactor; comfort managing small teams, including chairing stand-ups and developing detailed roadmaps for a 3–6 month research vision.

Nice To Haves

Experience with Kubernetes, preferred over SLURM.
Experience optimizing and profiling code on GPUs, such as writing CUDA kernels.

Responsibilities

Join an existing research workstream and lead projects.
Research conditions under which deceptive model behavior can be reliably detected and mitigated at scale, including large-scale transformer training.
Apply research insights to detect and mitigate vulnerabilities and other risks in frontier AI models.
Build frameworks and infrastructure that enable larger research questions and faster experimentation.
Depending on the area of expertise, contribute to transformer training, high-performance computing, distributed systems, or technical leadership and project planning.
Mentor other engineers or scientists in engineering skills.