Intern - AI - LLMs and Safety  

Position Overview:  We are seeking an enthusiastic and motivated AI Intern to join our innovative team. In this role, you will be at the forefront of enhancing the reliability, safety, and performance of AI models and systems. You will collaborate closely with AI researchers and product teams to contribute to the development of cutting-edge advancements in AI safety and responsible AI solutions.

Responsibilities:

• Assist in designing and implementing end-to-end safety-focused frameworks for LLMs.
• Implement risk mitigation techniques by building safe inference strategies.
• Identify vulnerabilities in AI systems and contribute to strategies for mitigation, including adversarial testing and bias detection.
• Collaborate with cross-functional teams to integrate safety mechanisms into AI workflows and pipelines.
• Stay up to date with the latest research papers, techniques, and advancements in deep learning and related fields.
• Strong software engineering and programming skills, and ability to quickly develop working prototypes from research ideas.

Requirements:

• Currently enrolled full time and pursuing Bachelor's, Master's, or Ph.D. degree in Computer Science, Electrical Engineering, or a related field.
• Must be graduating between December 2025 and June 2026.
• Available to work from May 27 - August 16 OR June 17 – September 6 – Summer 2025.

Job must-haves:

• Good understanding of natural language processing (NLP) fundamentals and techniques with a focus on architectures such as GPT, BERT, and their variants.
• Familiarity with distributed training and inference with large compute clusters.
• Familiarity with testing AI pipelines, data preprocessing, and model evaluation.

Great Candidates may have:

• Stellar publication record at top-tier conferences such as NeurIPS, ICLR, ICML, CVPR, ACL.
• Experience with state-of-the-art multimodal foundation models or a knowledge of inference optimization techniques for edge inference.
• Experience with advanced RAG pipelines.

Representative Projects (non-exhaustive):

• Design a safety pipeline for a new modality or for a specific customer workflow such as RAG.
• Compare safety standards of different architectural variants and LLM inference techniques.
• Scale and speed-up inference and design an end-to-end pipeline with a focus on safety on the edge.
• Build novel benchmarks and evals for safety.
• Design novel jailbreaking or defense techniques for red teaming/ safeguarding LLMs.
• Finetune or distill pre-trained LLMs on domain-specific datasets to improve task performance and reduce latency for safety workflows.

Compensation: starting from $50/hour for B.S.