Applied AI Research Intern

Applied AI Research Intern

We are an early-stage AI startup on a mission to enable the safe and compliant deployment of AI Agents in regulated industries, worldwide, through our Regulatory Guardrails & Auditing platform. We ensure conversational AI agents adhere to companies' business policies and regulations at scale, like Investment Guard. We envision a future where compliance is encoded in any autonomous system, steered by human experts. The goal is to augment compliance experts' capabilities, empowering them to become key enablers in the deployment of AI Agents in the most critical scenarios at scale. Co-founders Ari and Carlos come from leading ML platform at Twitter and LLM governance at Hugging Face. At Alinia, our Applied AI Research Interns play a pivotal role in exploring new possibilities and capabilities for Alinia's Alignment Platform. This role demands a scientific mindset, a technical understanding of LLMs, and LLM development experience. We expect our interns to be self-motivated, action-oriented and hands-on. During the internship, interns will be expected to design, execute, and deliver their solutions to key challenges our customers encounter with the safe and responsible deployment of LLM applications. However, we don't expect our interns to work alone. A mentor will be assigned throughout the internship to provide guidance, planning, and support. Interns will also be expected to participate in weekly meetings where they will share progress and present results. This role is designed for students who are ready to apply their expertise actively and decisively within a dynamic development environment.

Developing a robust and holistic alignment strategy and platform that combines state-of-the-art evaluation and optimization techniques, Responsible AI best practices and the realities of running a business is a complex task with applied research at the center. As an Applied AI Research Intern at Alinia, you will experimentally design, develop, evaluate, and execute a project to support the training, evaluation, or use of Large Language Model (LLM)-based guardrails. Example projects include, but are not limited to:

• The development of a robust, automated approach to quantitatively measure synthetically-generated data quality.
• The incorporation of explanations for our LLM guardrails to provide post-hoc rationales for why content was blocked.
• Pruning and quantization of LLMs without sacrificing out-of-distribution performance.

We strive to create intern projects that provide the right mix of problem solving, learning, and real-world application as possible while also aligning with the student's interests. As a member of a small team, this role presents a unique opportunity to make direct contributions to a real-world product. Your work will directly shape the art of the possible for our customers and their clients.

Minimum Qualifications:

• MS student in Computer Science, AI, Linguistics, or a related field.
• Proven experience in LLMs, ML, or NLP.
• At least one publication in reputable AI, ethics, or machine learning conferences and journals.
• Strong programming skills in Python.
• Experience with ML frameworks such as TensorFlow, PyTorch, or JAX.
• Demonstrated knowledge and practical experience in LLM training (Please note: this is a strict requirement for the position).

Preferred Qualifications:

• PhD student in Computer Science, AI, Linguistics, or a related field.
• Contributions to open-source projects or public datasets in the field of AI.
• First-author publications at peer-reviewed AI conferences.
• Experience with synthetic data generation, LLM as a judge frameworks, LLM post-training.
• Experience with explainable AI (XAI).

Why Join Alinia?

• Cutting-edge tech: Work on one of the most important challenges in AI—alignment, safety, and trust
• Flexible work: Hybrid or remote work, with preference for CET time zone
• Collaborative culture: Small, experienced, mission-driven team
• Impact: Directly shape the technical foundation of an AI governance platform adopted by enterprises