Research Intern

Sierra•San Francisco, CA

5d•Onsite

About The Position

Sierra’s Applied Research team works to improve the models behind its customer-facing AI agents, making them better, faster, and more cost-effective across many tasks. Its research includes post-training open-weight models for specialized capabilities, developing new approaches to voice AI, and advancing benchmarks used to evaluate agents, including τ-bench. The Applied Research Intern will join the team for a mentored research project focused on a meaningful problem in real-world AI agent performance. The intern will frame a research question, conduct rigorous experiments, and may write up findings for submission to a top research venue. The work may contribute both to the research community and to production agents. The internship is based in Sierra’s San Francisco office, with the team working in person Monday through Thursday. Duration and start dates are flexible.

Requirements

Currently pursuing a PhD in computer science, machine learning, or a related field.
A research track record, such as first-author publications or submissions to venues such as NeurIPS, ICML, ICLR, ACL, EMNLP, Interspeech, or ICASSP.
Strong experimental skills, including designing clean experiments, building careful baselines, and accurately interpreting results.
Fluency in Python and modern machine learning tools, such as PyTorch and LLM inference and training stacks.
High agency and a builder’s mindset, with the ability to work across the stack to get experiments running.
Clear, direct communication in writing and research presentations.
Currently enrolled in a PhD program.

Nice To Haves

Research experience with LLM agents, tool use, or multi-turn dialogue.
Experience designing benchmarks or evaluations, especially those tied to real-world outcomes.
Experience with supervised fine-tuning, reinforcement learning, or long-horizon optimization.
Experience with speech, voice, or multimodal models.
Open-source contributions to machine learning research code or benchmarks.

Responsibilities

Own a research project from framing the question and running experiments through writing and submitting a paper.
Advance agent evaluation, for example by extending τ-bench to new domains, modalities, or long-horizon tasks, or by developing new ways to measure reliability and consistency.
Study how agents reason, use tools, and follow policies over multi-turn conversations, and identify where and why they fail.
Explore post-training methods, such as supervised fine-tuning and reinforcement learning, to improve agent behavior and rigorously measure results.
Use large-scale proprietary datasets of text and audio interactions to identify research questions grounded in real customer problems.
Collaborate with Sierra researchers and engineers and, where appropriate, translate findings into production improvements.

Benefits

Flexible (unlimited) paid time off.
Medical, dental, and vision benefits for employees and their families.
Life insurance and disability benefits.
Retirement plan dependent on country of employment.
Parental leave.
Fertility and family-building benefits through Carrot.
Lunch, snacks, and coffee.
Discretionary benefit stipend.
Free alphorn lessons.
Eligible full-time employees can participate in Sierra’s equity plans, subject to applicable plan terms and policies.