Multimodal Engineer

Nori Robotics (YC S26) builds affordable humanoid robots. Our work connects some of the world's most novel multimodal, creative, and behavioral datasets to robots and consumer applications used by real customers. We're looking for hungry, talented, curious people to join our team.

Scope: Own the dataset-processing and quality pipeline for synchronized multicamera video, multichannel audio, and sensor recordings. Turn raw sessions into aligned, validated, searchable datasets that researchers can use for model development and evaluation.

What you will work on

• Build decoding, synchronization, calibration, normalization, and metadata pipelines for large video, audio, and sensor datasets.

• Implement QA/QC for corrupt or missing files, dropped frames, audio alignment, synchronization drift, and annotation consistency.

• Run existing models for captioning, tracking, embeddings, and automated quality checks; measure errors and route uncertain outputs for review.

• Version datasets and derived features with source lineage; deliver research-ready exports and improve processing throughput, recovery, and cost.

Requirements

• Strong Python and experience building large-scale media or sensor-data pipelines, including low-level formats, timing, and data integrity.

• Experience with batch or GPU inference, automated QA/QC, dataset versioning, and failure recovery on production datasets. A PhD is not required.

• Nice to have: camera calibration, signal processing, FFmpeg/GStreamer, or large-scale dataset search.

San Francisco, CA. Full time or contract. Primarily in person; remote arrangements are available.

Compensation. Contract: $70-$110 per hour. Full time: $130,000-$200,000 per year. Base pay varies with market location, job-related knowledge, skills, and experience.