aVenture · Jun 2026
Third-party company profile summarizing Hillclimb as a San Francisco startup founded in 2025 with about $500K raised and a focus on training data for recursive self-improvement.
The world's first human superintelligence community
Company research
Public company, workplace, funding, and market signals
Updated Jul 30, 2026
Overview
YC-backed San Francisco startup building specialized math-focused training data and reinforcement learning environments for frontier AI labs, aimed at improving model capabilities toward recursive self-improvement.
Primary product
Math-focused training data and RL environments for frontier AI labs
Founded
2025
Headquarters
San Francisco, California, United States
Team size
1-10
Industry
Software Development
Sub-industry
AI training data / reinforcement learning environments
Offices
San Francisco, California, United States
United Kingdom
Kazakhstan
Business model
B2B data sales
custom training-data services
RL environment design services
Funding
Stage
Pre-seed / YC-backed
Total raised
$500K
Latest round
Pre Seed · Oct 2025
Latest amount
$500K
Oct 2025 · Y Combinator
People & investors
Leadership
Founder
Founder
Founding Engineer
Research Engineer
AI Mathematics Consultant
Work & culture
Research-first, elite-talent culture centered on IMO medalists, Lean experts, and fast feedback loops; the company explicitly treats contributors as partners rather than gig workers and welcomes new grads for some roles.
Visa sponsorship
Limited
Compensation
Public hiring pages list Member of Technical Staff roles at $200K-$300K salary plus 1.00%-5.00% equity, and internship roles at $10K/month; new grads are explicitly welcome for some MTS roles.
Product & market
Pricing
Custom B2B project-based data sales / services to frontier AI labs
Differentiators
- —Focuses on recursive self-improvement rather than generic model hosting
- —Builds specialized reinforcement learning environments
- —Uses elite mathematicians and IMO/Putnam/Lean talent as the data source
- —Direct collaboration and co-authorship with contributors
- —Shortens feedback loops from months to hours
- —Sells data directly to frontier labs
Technology
Python
C++
Linux
Docker
Slurm
distributed systems
reinforcement learning
natural language processing
cloud infrastructure
CI/job orchestration
Customers
frontier AI labs
Nous Research
Competitors
traditional data vendors
general-purpose data labeling vendors
Scale & contact
Estimated monthly visits
2.5K
Latest news
Startuply.vc · May 2026
Profile piece frames hillclimb as a 2025 YC-backed startup building specialized training data and RL environments for autonomous research agents.
LinkedIn / Y Combinator · Nov 2025
YC LinkedIn post highlights hillclimb's elite math-talent community, faster feedback loops, and work with frontier labs.
Y Combinator · Nov 2025
YC launch describing hillclimb as a human-superintelligence community building math RL environments for frontier labs and referencing the Nous Research partnership.
LinkedIn · Oct 2025
Founder announcement that hillclimb was backed by Y Combinator and was building math RL environments for post-training.
Jobs and careers at hillclimb
0 jobs at hillclimb
No open positions
Check back later for new openings