Sieve — Member of Technical Staff, Applied Research

Type: Full-time | On-site | San Francisco, CA Compensation: $150,000–$350,000 + 0.05%–0.4% equity Hiring count: 4 Visa sponsorship: Yes — H-1B, O-1, OPT Reports to: TBD (CTO / CEO are in the loop per interview process)

Sieve is the only AI research lab exclusively focused on video data, combining exabyte-scale video infrastructure, novel video understanding techniques, and dozens of data sources to build datasets that push the frontier of video modeling. Video makes up ~80% of internet traffic and underpins creativity, communication, gaming, AR/VR, and robotics; Sieve targets the biggest bottleneck to progress in those domains — high-quality training data. The company has partnered with top AI labs and generated meaningful revenue last quarter with a team of just 12.

Founded: 2022 | Team size: ~12 (listed 11–50) | Total funding: Series A (amount undisclosed) Investors: Matrix Partners, Swift Ventures, Y Combinator, AI Grant Industry: AI Tools / Video data Website: sievedata.com Office: San Francisco, CA

• Frontier video AI: The only research lab working exclusively on video data, partnering with top AI labs at internet scale.
• Tiny, elite team: ~12 people, revenue-generating, recently raised a Series A from Tier 1 firms.
• Real ownership: Build research building blocks and pipelines that directly shape what downstream AI models and agents can do.

• Intake video on file; no transcript available to summarize. Re-confirm key points with David if needed before scoring.

As an applied research engineer, you'll build high-performance building blocks and large-scale pipelines to understand video with high precision at internet scale — frequently working on ambiguous research problems and finding clever techniques to solve them, across computer vision, audio processing, and text processing.

• →Build high-performance research building blocks and large-scale video understanding pipelines
• →Squeeze performance out of models + APIs via pre/post-processing, parallelism, pipelining, inference optimization, and occasional fine-tuning
• →Work across computer vision, audio processing, and text processing domains
• →Take problems from ambiguous research questions to production systems

Tech stack: Python, PyTorch (or similar ML frameworks)

• 3+ years of experience in computer vision or audio processing
• Strong Python developer with hands-on experience in PyTorch or similar ML frameworks
• Excellent communication skills with customers and external teams
• Write clean, maintainable code
• Deep passion for video domain and media technologies
• Motivated by building end-to-end products
• Break down problems from customer level impact to building blocks
• In-person at SF HQ

• Research background, able to work independently on problems
• Experience at top tech companies/research labs
• Direct experience in computer vision/robotics

• Less than 3 years of relevant experience
• Lack of research/independent work experience
• No background at top-tier tech/research companies

Salary$150,000–$350,000Equity0.05%–0.4%On-site policy In-person at SF HQ Visa sponsorshipH-1B, O-1, OPT Employment type Full-time Location San Francisco, CA

From the Required Candidate Q&A card — section was truncated behind "Show More"; only the first three captured. Re-paste expanded for the full list.

1. LinkedIn URL
2. Where are you currently based?
3. Are you willing to relocate to San Francisco? (If you already live here, click yes.)

Stage 1 — Initial Screen — Recruiter/early screen. Stage 2 — Technical Chat with CTO — Technical discussion with the CTO. Stage 3 — Technical Interview — Deeper technical evaluation. Stage 4 — Chat with CEO — Conversation with the CEO. Stage 5 — On Site — In-person on-site round. Stage 6 — Offer Extended Stage 7 — Candidate Hired — Candidate accepts and starts.

(Contrario pipeline also lists "Pending Approval" and "Application Review" as preceding platform/admin stages.)

Updated June 26, 2026 Top tech / research labs & robotics — Meta, NVIDIA, OpenAI, Google DeepMind, Waymo, Google, Intrinsic, David AI (audio datasets for speech & conversational AI)

For reference only — do not source these specific profiles. Marked "DO NOT CONTACT" on the role page. LinkedIn URLs were not exposed as text in the HTML (icon links only).

• Felipe Moreno
• Vahe (Vage) Taamazyan
• Gaurang Bharti
• Ben Caffee
• Geoffrey Bradway
• Jiqi Yang
• Jerry Ji
• Aun Siddiqui
• Jack Langerman
• Thanos Papadopoulos

• None provided on the role page. Activity log shows multiple candidates rejected by the company (no reasons given).