Deep Learning Engineer

About the role

You will own the design and implementation of cutting edge deep learning models that power real time perception for transit and government clients. This role requires you to translate ambiguous product requirements into robust technical solutions that directly impact customer outcomes and safety. You will be responsible for researching, prototyping, and scaling models that interpret complex real world visual data under strict operational constraints. You will work at the intersection of research and production to ensure that Hayden AI's capabilities remain at the leading edge of computer vision. Your work will directly influence how cities understand and manage mobility, making transportation systems more efficient and safer for the public. You will partner with product managers and domain experts to ensure that your models solve the hardest problems faced by our customers. Finally, you will mentor junior engineers and contribute to the technical direction of the deep learning organization.

Key facts

What you'll do

Architect and develop 3D vision models that estimate depth and structural understanding of urban environments from mobile sensor platforms.

Design and implement video and temporal behavior models that predict intent and movement patterns for vehicles and pedestrians in complex traffic scenarios.

Lead experiments with Vision Language Models, adapting and fine tuning them for specialized perception tasks in transit and public safety contexts.

Evaluate and benchmark foundation models in perception, selecting and customizing architectures that meet latency, accuracy, and robustness targets.

Collaborate closely with the platform team to integrate Nvidia edge device stack knowledge, optimizing CUDA kernels and kernels configuration for maximum throughput on embedded hardware.

Own the end to end model lifecycle from data exploration and labeling strategy, through training, validation, and deployment in production cloud and edge environments.

Partner with cross functional stakeholders to define product metrics, validate model performance in the wild, and drive continuous improvement based on field data.

Implement scalable MLOps pipelines for experiment tracking, model versioning, and automated retraining to support rapid iteration and reliability.

Conduct rigorous performance analysis, debugging failed cases, and proposing data-centric or architectural improvements to perception systems.

Represent Hayden AI in technical discussions with customers and partners, explaining model capabilities, limitations, and trade offs in clear and actionable terms.

Contribute to hiring and knowledge sharing by conducting code reviews, documenting best practices, and mentoring other engineers on the team.

Participate in on call rotations to support deployed models, triage issues, and guide incident response efforts when perception failures occur.

Drive research initiatives that explore novel model architectures, loss functions, and training strategies to maintain a competitive edge in mobile perception.

Act as a technical leader in defining standards for model interpretability, safety validation, and compliance with transportation regulations.

Requirements

You have 1 to 2 years of hands on experience building and deploying machine learning models for perception in production settings.

You possess solid hands on experience designing, training, and evaluating machine learning models specifically for perception tasks in dynamic environments.

You can independently design and maintain machine learning pipelines, from data ingestion and labeling to model training, validation, and deployment on major cloud platforms such as AWS, GCP, or Azure.

You are familiar with MLOps best practices, including experiment tracking, model versioning, and automated workflows that ensure reproducibility and reliability.

You have strong programming skills in PyTorch and Python, with a proven track record of writing clean, maintainable, and efficient code.

You are a good communicator who works effectively in a fast paced startup environment, collaborating closely with cross functional teams.

You demonstrate the ability to learn quickly, adapt to changing priorities, and take ownership of ambiguous problems without direct supervision.

You hold a Bachelor's or Master's degree in Computer Science or a closely related technical field.

You have experience working on perception problems in self driving car companies that align closely with the mission and technical challenges faced by this team.

Practical notes

This role follows a hybrid schedule with at least 3 days in-office per week, based in San Francisco.