Frontier AI Research Intern: Agentic Systems

Together; San Francisco (CA); On-site; USD 80,000 - 96,000; Full time

Benefits offered by this job

Housing stipend

Competitive benefits

Job summary

T Together AI is seeking a research intern to explore frontier agentic AI—aligning, scaling, and evaluating models for complex, multi-step tasks in text and speech. You will work with colleagues across ML, NLP, and systems to push the boundaries of agentic behavior and reliability.

The internship runs 12 to 14 weeks in San Francisco, with compensation including housing stipends and competitive benefits. You will publish findings and contribute to ongoing research on self-learning and

Qualifications

• Ph.D. candidate in CS/EE/Info Science or related field.
• Publications at top ML/NLP conferences (NeurIPS, ICML, ICLR, ACL, EMNLP, Interspeech).
• Strong knowledge of ML/DL fundamentals.
• Experience with DL frameworks (PyTorch, JAX).
• Understanding of how LLMs work.
• Strong Python programming skills.
• Familiarity with Transformer architectures and foundation models.

Responsibilities

• Research and implement novel techniques in frontier AI focus areas.
• Design and conduct rigorous experiments to validate hypotheses.
• Document findings in scientific publications and blog posts.
• Communicate project plans, progress, and results to the team.

Skills

Python

ML basics

Deep Learning

Transformer architectures

Education

Ph.D. candidate in Computer Science / Electrical Engineering / Information Science

Tools

PyTorch

JAX

Transformers

Job description

T Together AI is seeking a research intern to explore frontier agentic AI—aligning, scaling, and evaluating models for complex, multi-step tasks in text and speech. You will work with colleagues across ML, NLP, and systems to push the boundaries of agentic behavior and reliability.

The internship runs 12 to 14 weeks in San Francisco, with compensation including housing stipends and competitive benefits. You will publish findings and contribute to ongoing research on self-learning and