6+ years of published research - so Rockfish synthetic data is statistically grounded, not guesswork
Rockfish is built on peer-reviewed work from Carnegie Mellon on time-series data generation, rare event synthesis, and privacy-preserving generative models — published at NeurIPS, ICML, AAAI, SIGCOMM, and IMC. The science is what makes the synthetic data trustworthy.
Since 2016, at Carnegie Mellon, co-founders Giulia Fanti and Vyas Sekar have made significant advances in the theory and practice of generative models such as developing state-of-art models for timeseries data, improving the stability and privacy of generative algorithms, practical approaches for rare sample generation, and domain-specific adaptations of deep generative models (e.g., telecommunications, IoT). These peer-reviewed research papers have appeared at prestigious AI/ML venues such as NeuRIPS, ICML, AAAI, and domain-specific venues such as IMC, SIGCOMM.
Latest Research 2026
AgentFuel: Expressive Evals for Time-Series Data Analysis Agents
Popular LLM-based analytics agents fail on domain-specific, incident-aware time-series questions — because they've never been evaluated on realistic domain data. AgentFuel is a framework for generating expressive, grounded evaluation suites from your actual time-series data, so you can measure and fix agent failures before they reach users.
Read more about some of these research foundations below.
Rare Classes
GAN-generated Samples