Founding AI Research Engineer

U.S. citizenship is required for this position.

Candidates must be eligible to obtain and maintain a U.S. government security clearance.

An active Top Secret/SCI clearance is preferred but not required.

PREDICTION is building advanced AI systems for national security.

We are looking for an exceptional Founding AI Research Engineer to join our core technical team and help define the AI research direction of the company.

This is a hands-on research and engineering role for someone who wants to work on difficult, real-world AI problems across video, imagery, text, sensors, structured data, and other signals.

We are not looking for someone who simply connects existing APIs.

We are looking for someone who can understand difficult machine learning problems, design new approaches, run rigorous experiments, build models, and turn successful research into systems that operate in the real world.

The level is flexible. We are interested in exceptional candidates ranging from outstanding early-career researchers to Senior, Staff, and Principal-level engineers.

What You’ll Work On

• Multimodal AI and reasoning

• Computer vision and video understanding

• Vision-language models and foundation models

• Temporal reasoning and long-term context

• Sensor fusion and multimodal data fusion

• Behavioral understanding and anomaly detection

• Object detection, tracking, and representation learning

• Agentic AI systems

• Model evaluation, calibration, robustness, and uncertainty

• Training, fine-tuning, post-training, and optimization

• AI systems operating in secure, edge, and mission-critical environments

What You’ll Do

• Research and develop new AI capabilities from idea through deployment

• Design and test new multimodal architectures and approaches

• Train, fine-tune, benchmark, and evaluate deep learning models

• Build systems that reason across multiple types of data simultaneously

• Develop methods for video understanding and temporal reasoning

• Implement and evaluate ideas from current AI research

• Design rigorous experiments and evaluation methodologies

• Investigate model failures and improve reliability, accuracy, and robustness

• Rapidly prototype new ideas and turn successful research into production systems

• Work closely with engineers and national security professionals on real operational problems

• Help define PREDICTION’s long-term AI research and technical direction

Who We’re Looking For

We care more about exceptional technical ability than titles or years of experience.

You may be completing a PhD.

You may already be a senior researcher or engineer.

You may have spent years publishing research, or you may have spent those years building exceptional systems.

What matters is what you can build, discover, and demonstrate.

Strong candidates will have deep experience in one or more of the following:

• Machine learning and deep learning

• Multimodal learning

• Computer vision

• Video understanding

• Vision-language models

• Transformers and foundation models

• Temporal modeling

• Representation learning

• Sensor fusion

• Agentic systems

• Reinforcement learning or post-training

• Model evaluation and calibration

You should have strong Python skills and significant experience with modern deep learning frameworks such as PyTorch.

You should also be comfortable reading research papers, understanding the underlying ideas, implementing them, identifying weaknesses, and developing better approaches.

What Will Get Our Attention

We want to see evidence of exceptional technical work.

That could be:

• Research published at NeurIPS, ICML, ICLR, CVPR, ICCV, ECCV, ACL, EMNLP, AAAI, KDD, or another leading venue

• A model or ML system you designed and built

• An exceptional open-source project

• A technically difficult system deployed in the real world

• Original research with meaningful results

• A GitHub repository that demonstrates exceptional engineering ability

• Experience at a leading AI lab, technology company, national laboratory, or advanced research organization

A PhD is valuable, but it is not required.

Why PREDICTION

At PREDICTION, you will not be one researcher among hundreds.

You will join a small technical team where your work can directly influence the models, architecture, research agenda, product, and direction of the company.

PREDICTION is a New York-based U.S. technology company building advanced AI systems for national security.

The company was built alongside senior leaders from the U.S. federal government, Intelligence Community, and law enforcement, together with engineers and researchers from leading technology and academic institutions.

We are looking for exceptional people who want significant technical ownership and the opportunity to solve problems that matter outside of a benchmark.