PhD Intern, Research - Model Alignment

Description

Are you excited about advancing the next generation of intelligent design systems? Do you want to explore how reinforcement learning can make generative models more adaptive, diverse, and aligned with human creativity? In this role you will join Autodesk’s Research team to advance reinforcement learning for generative design models, develop and evaluate new algorithms that enable models to self‑improve, learn from feedback, and generalize across diverse tasks. You will prototype and test ideas that could impact Autodesk products such as Fusion and AutoCAD and generate publishable insights for the broader research community.

The intern will conduct research on self‑play reinforcement learning methods for structured generative tasks and design RL‑based alignment frameworks using PyTorch, Ray, and RayLightning for distributed training. You will prototype algorithms that improve the validity, diversity, and stability of generated results, and collaborate with Autodesk Research scientists and engineers to integrate learned models into real‑world generative workflows. Findings will be documented and may contribute to peer‑reviewed publications at venues such as NeurIPS, ICLR, or ICML.

• Conduct research on self‑play reinforcement learning for structured generative tasks.
• Design and implement RL‑based alignment frameworks using PyTorch, Ray, and RayLightning for distributed training.
• Prototype algorithms to improve validity, diversity, and stability of generated results.
• Collaborate with Autodesk Research scientists and engineers to integrate models into real‑world workflows.
• Document findings and contribute to peer‑reviewed publications (e.g., NeurIPS, ICLR, ICML).

Requirements

Candidates must be currently pursuing a PhD in Computer Science, Machine Learning, Artificial Intelligence, or a related field. A solid background in reinforcement learning, deep learning, language models, and generative modeling is required, along with proficiency in Python, PyTorch, and deep RL frameworks. Understanding of distributed training or experience with Ray/RayLightning, strong analytical abilities, and effective communication skills are also essential.

Prior research experience in self‑play RL, model alignment, or structured generative modeling (e.g., text, geometry, or graph domains) is highly valued. Familiarity with CAD representations and graph neural networks, as well as publication or preprint experience in top‑tier conferences such as NeurIPS, ICLR, ICML, CVPR, or RLC, will strengthen an application. Candidates should be enthusiastic about bridging research and real‑world generative design systems and able to work independently while collaborating across multidisciplinary teams.

Autodesk’s software powers the creation of greener buildings, cleaner cars, smarter factories, and blockbuster movies, helping innovators turn ideas into reality. The company’s culture—centered on collaboration, customer focus, and societal impact—guides how employees work, connect, and show up in the world. As an Autodesker you can do meaningful work that builds a better world designed and made for all.

• Pursuing a PhD in Computer Science, ML, AI, or a related field with a strong RL and generative modeling background.
• Proficient in Python, PyTorch, and deep RL frameworks; experienced with distributed training (Ray/RayLightning).
• Strong analytical and communication skills for interpreting experimental results.
• Preferred: prior research in self‑play RL, model alignment, or structured generative modeling; familiarity with CAD and graph neural networks; publications in top‑tier conferences.

Benefits

The 2026 Canada program runs for 16 weeks from May 4 to August 21 and is a paid internship. Interns will contribute to meaningful projects, receive mentorship from industry leaders, and participate in tech talks and activities designed for personal and professional development. Autodesk’s Flexible Workplace approach supports office, remote, and hybrid work preferences.

Salary is part of Autodesk’s competitive compensation package and offers are based on experience, education level, and geographic location.

• 16‑week paid Canada internship (May 4 – August 21 2026) with mentorship, tech talks, and flexible workplace options.
• Autodesk offers competitive compensation, a culture of belonging, and opportunities to work on impactful design software.

Training + Development

Interview process

Visa Sponsorship

Security clearance