About Sieve
Sieve works with frontier AI labs and enterprise customers on highly specific dataset problems — building custom algorithms, models, and data pipelines at scale. The company focuses on video, audio, and multimodal data processing for AI training and evaluation. The team is four people and moving fast to scale deployments across leading AI labs.
About the Role
You'll own end-to-end dataset projects for customers, from untangling ambiguous requirements through shipping production systems that find, generate, filter, transform, evaluate, and package high-quality datasets. This is a high-agency role working directly with customers and internal teams, combining research prototypes with reliable production pipelines. You'll ship fast, move between different technical domains within each project, and own customer outcomes directly.
What You'll Own
- Work directly with customers to translate ambiguous dataset needs into concrete technical systems and delivery timelines
- Build custom algorithms, models, and large-scale data pipelines spanning computer vision, audio processing, text processing, and metadata analysis
- Move between research prototypes and production systems, using models and APIs creatively
- Break down customer-level goals into models, heuristics, infrastructure, and QA steps
- Optimize performance through pre/post-processing, parallelism, inference optimization, fine-tuning, and evaluation loops
Requirements
Must-Have
- Strong Python developer with hands-on experience building custom algorithms, model workflows, or large-scale data pipelines
- Strong intuition for dataset quality, filtering, labeling, evaluation, and edge cases
- Comfortable working directly with customers or external teams
- Experience with PyTorch or similar ML frameworks
Nice-to-Have
- Experience with large-scale video, audio, or multimodal data processing
- Active contributor to open-source projects
- Computer vision, audio, or video domain depth
Job Details
- Salary: $150,000 – $250,000
- Visa: H-1B, OPT sponsored
- Location: San Francisco — in-person
- Headcount: 4 open seats
- Equity: Competitive
]]>