Technical Staff - Applied Research & PyTorch (San Francisco)

Location: San Francisco, CA (In-Person)

Compensation: $150K-$350K Base + Equity + Benefits

Stack: Python, PyTorch, computer vision, multimodal AI, video understanding, audio processing, large-scale data pipelines, model optimization, and distributed inference systems.

TLDR

• Our client is building foundational infrastructure and datasets for the next generation of video AI, operating at the intersection of computer vision, multimodal learning, and large-scale data systems.
• The company has achieved meaningful commercial traction with a small, highly technical team while partnering with leading organizations pushing the frontier of AI research.
• Engineers work on some of the hardest problems in AI data: understanding, processing, and improving video at internet scale across vision, audio, and text modalities.
• This is a highly applied research role where success comes from shipping systems that improve model performance, dataset quality, and production outcomes—not publishing papers.
• You’ll have significant ownership, direct impact on core technology, and the opportunity to help shape a category-defining company in one of AI’s fastest-growing domains.

Requirements

• Strong Python skills and hands-on experience with PyTorch or comparable machine learning frameworks.
• Experience building applied machine learning systems in computer vision, audio processing, multimodal AI, or closely related domains.
• Experience working with models in production, including inference optimization, evaluation, experimentation, and performance tuning.
• Ability to break complex customer or product problems into scalable technical solutions and deliver end-to-end outcomes.
• Strong communication skills and a willingness to work closely with customers, researchers, and engineering teams in a fast-moving environment.

Bonus Skills

• Experience with video understanding, media processing, content intelligence, or multimodal AI systems.
• Experience improving model performance through data engineering, preprocessing, post-processing, fine-tuning, or evaluation techniques.
• Open-source contributions, technical side projects, or a demonstrated track record of building beyond day-to-day work responsibilities.
• Experience at an early-stage startup or other highly autonomous, high-ownership environment.
• Familiarity with large-scale data pipelines, distributed inference, or high-throughput AI infrastructure.

Responsibilities

• Design and develop scalable systems that improve video understanding and multimodal AI capabilities at large scale.
• Build data pipelines, evaluation systems, and infrastructure that support model development and dataset quality.
• Optimize model performance through experimentation, inference improvements, and creative approaches to challenging research problems.
• Collaborate with researchers, engineers, and customers to translate real-world needs into practical AI solutions.
• Drive technical decisions across data processing, model workflows, and production AI systems.
• Contribute to engineering excellence through high-quality code, strong technical execution, and continuous iteration.

About

• Our client is focused on solving one of the most important bottlenecks in modern AI: creating high-quality data and infrastructure for video understanding at scale. Video represents the majority of digital content and is increasingly central to applications across AI, robotics, media, and immersive computing.
• Applied Research Engineers sit at the center of the company’s competitive advantage, building systems that directly improve model performance, dataset quality, and customer outcomes.
• This is a hands-on role that combines research, engineering, and product impact, offering significant ownership and exposure to cutting-edge multimodal AI challenges.
• The role provides deep exposure to computer vision, audio processing, multimodal systems, large-scale AI infrastructure, and real-world machine learning applications operating at internet scale.