AI Research Engineer

About Poseidon

Poseidon is building the data infrastructure the next generation of AI will depend on. Foundation models are not limited by compute. They are bottlenecked by rare, high-quality, IP-safe data that actually improves performance: the long-tail, edge-case, multi-modal datasets that cannot be scraped or synthetically generated.

We are creating a data layer connecting AI companies with the datasets they need. Poseidon is the infrastructure that makes scalable, compliant, demand-driven data sourcing possible. Backed by a16z, we are early, moving fast, and looking for mission-driven teammates to shape this category.

Examples of the datasets we work with include conversational audio, complex video, domain-specific imagery, and other real-world data that enables models to perform reliably outside controlled environments.

The Role

We are hiring an AI Research Engineer to help advance Poseidon’s research in voice and multimodal machine learning.

This role sits at the intersection of machine learning research and applied systems work. You will collaborate closely with Poseidon researchers and engineers to design new models, run experiments on large-scale datasets, and publish research that advances the state of the art in voice and multimodal AI.

Voice modelling is currently the highest priority area, with video and other multimodal research emerging as the platform expands.

You will work closely with the research team and collaborate with Prof. Sandeep Chinchali (UT Austin, Stanford CS PhD), alongside engineers and researchers across the Poseidon platform.

What You'll Work On

• Research, develop, and tune voice foundation models, including systems trained on large-scale multilingual audio datasets.

• Explore multimodal AI systems combining voice, video, and language signals.

• Design and run experiments on large-scale training datasets across audio, video, and text.

• Publish research at leading conferences such as NeurIPS, ICML, ICLR, CVPR, and Interspeech.

• Collaborate with engineers to translate research prototypes into production machine learning systems.

• Help shape Poseidon’s research roadmap across voice and multimodal AI.

Research Environment

Poseidon supports publishing and academic collaboration. Our research work is designed to both advance the field and translate directly into production systems used by AI companies and enterprises.

This role is primarily research-focused, with roughly 80% of time spent on research and experimentation and approximately 20% on applied engineering work as research moves toward deployment.

Requirements

• Strong background in machine learning and deep learning; MS or PhD preferred.

• Experience working with voice, speech, or audio machine learning systems.

• Strong Python programming skills, including experience with PyTorch and core Python-based machine learning tooling such as Jupyter Notebook.

• Ability to design and run machine learning experiments at scale.

• Interest in publishing research and contributing to the broader ML community.

Preferred Background

• Experience with foundation models, speech models, or multimodal machine learning systems.

• Publications at major ML conferences (NeurIPS, ICML, ICLR, CVPR, Interspeech) or demonstrated interest in publishing research.

• Experience training models on large datasets or building machine learning systems in production.

• Familiarity with modern model training environments and infrastructure.

Team & Mentorship

You will work with a team of researchers and engineers focused on advancing AI systems that rely on high-quality, real-world training data.

This includes collaboration with Prof. Sandeep Chinchali, whose work spans machine learning systems and applied AI research, alongside engineers experienced in building large-scale production AI systems.