
Sepal AI empowers ambitious AI teams to build faster and safer foundational AI models by providing expert-grounded data at every stage of development. They focus on accelerating model iteration and ensuring quality through rigorous vetting and a vast pool of specialized PhD-level experts. Their services include providing novel training datasets, custom benchmarks, pre-launch testing, and agentic applications, all supported by a human-centric approach to AI development. Sepal AI is SOC 2 Certified and is relied upon by top AI labs that prioritize responsible and high-impact AI.

Sepal AI empowers ambitious AI teams to build faster and safer foundational AI models by providing expert-grounded data at every stage of development. They focus on accelerating model iteration and ensuring quality through rigorous vetting and a vast pool of specialized PhD-level experts. Their services include providing novel training datasets, custom benchmarks, pre-launch testing, and agentic applications, all supported by a human-centric approach to AI development. Sepal AI is SOC 2 Certified and is relied upon by top AI labs that prioritize responsible and high-impact AI.
What they do: Expert-grounded training data, RL environments, evaluation tooling, and model fine-tuning for frontier AI labs and enterprises
Headquarters: San Francisco, California, United States
Founders: Fedor Paretsky; Kat Hu
Stage / funding: Pre-Seed; raised a round announced Sep 25, 2024
Security / compliance: SOC 2 Certified
| Company |
|---|
Training data quality, model evaluation, and safety for large and frontier AI models
2024
Data and Analytics
500000 USD
“Y Combinator (lead/backing investor)”
Sepal is conducting a national research initiative to evaluate and benchmark AI systems for safety, reliability, and real-world technical capability. We are seeking experienced database administrators (DBAs)
to help define professional standards in data architecture, database performance, security, and maintenance. Your expertise will directly influence how future AI models are assessed and validated for mission-critical data operations.
Sample of Reported Job Titles
Database Administrator -- SQL Database Administrator -- Database Engineer -- Data Infrastructure Engineer -- Systems Database Specialist
MUST-HAVE QUALIFICATIONS
-- 4+ years full-time experience as a database administrator or in a closely related data infrastructure role
-- English language fluency
-- U.S. residency
-- Access to a laptop or desktop computer with reliable internet
-- Technical proficiency with relational databases (e.g., SQL Server, MySQL, PostgreSQL, Oracle) and familiarity with performance tuning, backup/recovery, security, and monitoring
-- Ability to follow detailed written instructions and complete online tasks independently
COMPENSATION AND LOGISTICS
-- Hourly pay: $38–$87 per hour, with potential task-based bonuses depending on experience and fit
-- Paid bi-weekly based on tracked and approved hours
-- Fully remote participation; work whenever you like as long as deadlines are met
-- Current employment in the field is not required
This is a short-term project
requiring roughly 40–100 hours of work total, between now and December 29th
. The engagement begins with a paid 2–4 day trial
.
WHAT YOU WILL DO
-- Complete a short professional questionnaire to outline your DBA background and technical skillset
-- Review AI-generated database queries, schema designs, troubleshooting procedures, and optimization recommendations for accuracy, safety, and feasibility
-- Help design realistic and edge-case scenarios involving data corruption recovery, query optimization, indexing strategies, replication, and security hardening
-- Provide expert insight into what high-quality versus low-quality database administration looks like in production environments
WHY PARTICIPATE?
Your real-world DBA expertise is essential for evaluating the safety and capability of next-generation AI systems. By participating, you directly contribute to building rigorous evaluation standards while being compensated for your knowledge. Top contributors may be invited to future research cohorts or advisory roles in ongoing AI capability and safety assessments.
LIMITED AVAILABILITY
We are selecting a limited number of experienced database administrators for this cohort. Applications are reviewed on a rolling basis, and slots will fill quickly.