[Remote] Research Intern (LLM)

Note: The job is a remote job and is open to candidates in USA. Abaka AI is focused on advancing artificial intelligence research, and they are seeking a Research Intern to contribute to the development of challenging QA datasets and evaluate large language models. The role involves collaboration with global researchers and requires strong analytical and execution skills. Responsibilities Design and construct high-quality, sufficiently challenging QA datasets (graduate/PhD level) inspired by GPQA, HLE, and AI4Sci families, collaborating with a global network of talented researchers Evaluate large language models on reasoning, factuality, and problem-solving benchmarks Develop review pipelines and quality-control criteria for expert-level question generation Analyze model outputs, conduct error taxonomy studies, and summarize insights for internal reports and research papers Collaborate with the 2077AI Foundation’s open-source benchmark teams on public dataset releases Skills Strong background in computer science, data engineering, artificial intelligence, or related fields, with hands-on experience in large-scale data systems 1+ years of experience with LLMs, prompt engineering, and evaluation frameworks (e.g., LM Eval Harness, OpenCompass) Excellent written and verbal English skills and analytical reasoning Strong execution and team management skills—able to translate high-level objectives into actionable plans and drive team outcomes Experience with formal methods, chain-of-thought evaluation, or curriculum generation Relevant publications in top conferences Company Overview Abaka AI is a leading AI company and we are committed to becoming the data partner in artificial intelligence industry. It was founded in 2021, and is headquartered in Palo Alto, California, USA, with a workforce of 51-200 employees. Its website is Company H1B Sponsorship Abaka AI has a track record of offering H1B sponsorships, with 2 in 2025. Please note that this does not guarantee sponsorship for this specific role.