Vāyuh

Rethinking Extreme Weather Intelligence

∂ρ/∂t + ∇·(ρv) = 0
∇²φ = ρ/ε₀

Rethinking
EXTREMEWEATHER
INTELLIGENCE

15
Years of Research
Since 2009
100x
faster than
physics models
4+
Catastrophes
Explore Products

Trusted By Industry Leaders

Britt Insurance
Lloyd's Lab
Aon
Emergent
UC Berkeley
Britt Insurance
Lloyd's Lab
Aon
Emergent
UC Berkeley
OUR APPROACH

Help Insurance understandCatastrophe

Vayuh combines atmospheric physics with modern AI to deliver operational forecasts, risk alerts, and explainable confidence. Our hybrid models are engineered with scientists and engineers from the UC Berkeley community and designed to turn weather into action—reliably and at scale.

  • Physics-informed AI — data assimilation, ensemble learning, and domain constraints for fidelity and lead time.
  • Catastrophe modeling — severe convective storms, hurricanes, floods, and other insurance-critical perils.
  • Operational by design — APIs, observability, and uncertainty bands to support decisions and SLAs.

Built by UC Berkeley scientists · Partners & collaborators: NSF · Oak Ridge · LBL

Physics + AI

Hybrid models blend atmospheric dynamics with machine learning for robust, high-resolution forecasts.

Catastrophe focus

Purpose-built for insurance catastrophes — hail, tornadoes, hurricanes, and severe convective storms that drive losses.

Operational reliability

API-first delivery, monitoring, and traceable runs to support enterprise workflows and SLAs.

Explainable confidence

Uncertainty bands, feature attributions, and scenario testing to communicate risk—not just a number.

Data assimilation
Peer-reviewed methods
Uncertainty quantification
MARKET INSIGHT

The Paradigm Shift in Insurance

Severe Convective Storms are now the primary driver of catastrophic losses

The New Reality

58%
of 2023 insurance losses

Severe Convective Storms (hail, tornadoes, derechos) caused $58 billion in insured losses in 2023, eclipsing hurricanes for the first time in history.

Losses increasing 5–7% annually

Why Models Fail

  • Physics models can't capture extreme convection
  • Traditional cat models use 30‑year‑old statistics
  • Weather AI optimizes for averages, not extremes

Result: 40% of SCS losses occur outside modeled high‑risk zones

Market Opportunity

10xBetter Predictions
  • Write profitable business in “no–go” zones
  • Optimize reinsurance with granular risk views
  • Price 30–40% more accurately than competitors

* Based on industry reports and market analysis

OUR PRODUCTS

Insurance-Ready Risk Intelligence

From severe storms to portfolio analytics — purpose-built solutions for modern insurers.

PORTFOLIO EXPOSUREGEOGRAPHIC DISTRIBUTIONSCS INTELLIGENCEREAL-TIME ANALYSISSeverityHIGHImpact45MTime2HRISK ANALYTICSTotal Exposure$2.3BAt Risk$387MCoverage92%
SEVERE CONVECTIVE STORM SUITE

SCS Risk Intelligence

The most comprehensive severe storm forecasting and risk assessment platform. Now the #1 driver of catastrophic losses.

  • Hail Forecasting – Size, intensity, accumulation zones
  • Tornado Risk – Path prediction with STP-awareness
  • Wind & Derecho – Straight-line wind event modeling
  • Real-time Alerts – 15-min to 14-day forecasts
HURRICANE RISK PLATFORM

Hurricane Intelligence

Physics-informed AI for hurricane track, intensity, and rainfall predictions with uncertainty quantification.

  • Track Forecasting – Ensemble predictions to 10 days
  • Rapid Intensification – 6-24hr RI probability
  • Precipitation – Rainfall accumulation mapping
  • Storm Surge – Coastal inundation modeling
PORTFOLIO INTELLIGENCE

Portfolio Risk Management

Transform your exposure data into actionable intelligence with real-time portfolio monitoring.

  • Exposure Management – Real-time portfolio views
  • PML Analytics – Dynamic probable maximum loss
  • Risk Layering – Optimize retention & reinsurance
  • Early Warning System – Portfolio-specific alerts
DATA PLATFORM

Data Services & APIs

Access our models and datasets through flexible APIs or direct data feeds.

  • REST APIs – Real-time weather & risk data
  • Event Catalogs – Historical & stochastic events
  • Hazard Layers – High-resolution peril maps
  • Platform Integration – Seamless workflow integration

Research Excellence & Innovation

20k+ Citations
ACM Gordon Bell Prize 2018
Published Deep Learning Authors
Patent-Pending Methods
20k+ Citations
ACM Gordon Bell Prize 2018
Published Deep Learning Authors
Patent-Pending Methods
01
20k+ Citations

20k+ Citations

15+ Years of pioneering research studying the atmosphere using AI/ML. Our research has been cited over 20,000 times

02
ACM Gordon Bell Prize Winners

ACM Gordon Bell Prize Winners

Recognized with the prestigious ACM Gordon Bell Prize in 2018 for outstanding achievement in high-performance computing

03
Published Authors

Published Authors

Co-authors of “Deep Learning for Earth Science” (2021), advancing the field of AI applications in climate science

Vayuh isn't just improving forecasts—we're re-architecting catastrophe risk management for a more volatile world.

Mayur Mudigonda

Founder & CEO, Vāyuh

Insurance & Asset Managers

Reduce tail risk while lifting precision

Physics‑AI risk models quantify peril exposure at site and portfolio levels—improving underwriting and capital allocation.

Climate change has transformed the risk landscape—historical “1 in 100-year” events now occur with increasing frequency and intensity. Our AI models combine future-looking physics-based forecasts with machine learning to generate accurate, high-resolution risk maps for wildfire, extreme wind, temperature, precipitation, and severe convective weather.

Property RiskPortfolio RiskCatastrophe RiskClimate Risk
HAILTORNADOWINDLIGHTNINGGLOBAL RISK ASSESSMENT
HOW WE DO IT

From Data → Decisions

We combine atmospheric data with an extremes–first foundation model, specialize for key tasks, and deliver insurance–ready products.

DataAI ModelInsightsProducts
1

Atmospheric Data

ERA5
Global reanalysis
HRRR
High-res rapid refresh
NCEI
Climate archives
2

Vayuh Catastrophe Foundation Model

Climate Model
Physics-AI hybrid engine
ClimaXFourCastNetGraphCastPangu-Weather
3

Task Specialization

Downscaling
Weather Forecasting
Climate Assessment
Insurance Layers
4

Insurance Products

Extreme Weather Risk Assessment
Hazard Layers
Portfolio Analytics

Meet Our Team

Scientists and engineers building physics‑first AI for weather and risk.

Mayur Mudigonda, Ph.D.
Mayur Mudigonda, Ph.D.
Founder & CEO
Berkeley Artificial Intelligence Research
Berkeley Lab (LBL, DOE) Researcher
Pratik Sachdeva, Ph.D.
Pratik Sachdeva, Ph.D.
Scientist, Engineer
UC Berkeley
Physics, AI, Neuroscience
Farid Jiandani
Farid Jiandani
Jiandani Family Fund
Andy Konwinski
Andy Konwinski
Investor
Co-Founder Databricks
Co-Founder Perplexity
Prabhat Ram, Ph.D.
Prabhat Ram, Ph.D.
Advisor
Principal, Microsoft Azure
Berkeley National Lab, UC Berkeley
Olivier Collignon
Olivier Collignon
Advisor
Director of Product, RMS
Cape Analytics
Bharath Madhusudhan
Bharath Madhusudhan
Investor
Co-Founder Securly
Chandini
Chandini
Business Development
Ex 500 Startups
Caterpillar
Sabbih
Sabbih
AI Research Engineer
Zayd Enam
Zayd Enam
Investor
Co-Founder Cresta.ai
Ashesh Chattopadhyay, Ph.D.
Ashesh Chattopadhyay, Ph.D.
Assistant Professor, UCSC
Advisor
Abhimanyu
Abhimanyu
Senior Software Engineer
Blake Tickell
Blake Tickell
AI Research Engineer
Sean Spain
Sean Spain
Head of Product/GTM
Brit, Ki, Deloitte
Mayur Mudigonda, Ph.D.
Mayur Mudigonda, Ph.D.
Founder & CEO
Berkeley Artificial Intelligence Research
Berkeley Lab (LBL, DOE) Researcher
Pratik Sachdeva, Ph.D.
Pratik Sachdeva, Ph.D.
Scientist, Engineer
UC Berkeley
Physics, AI, Neuroscience
Farid Jiandani
Farid Jiandani
Jiandani Family Fund
Andy Konwinski
Andy Konwinski
Investor
Co-Founder Databricks
Co-Founder Perplexity
Prabhat Ram, Ph.D.
Prabhat Ram, Ph.D.
Advisor
Principal, Microsoft Azure
Berkeley National Lab, UC Berkeley
Olivier Collignon
Olivier Collignon
Advisor
Director of Product, RMS
Cape Analytics
Bharath Madhusudhan
Bharath Madhusudhan
Investor
Co-Founder Securly
Chandini
Chandini
Business Development
Ex 500 Startups
Caterpillar
Sabbih
Sabbih
AI Research Engineer
Zayd Enam
Zayd Enam
Investor
Co-Founder Cresta.ai
Ashesh Chattopadhyay, Ph.D.
Ashesh Chattopadhyay, Ph.D.
Assistant Professor, UCSC
Advisor
Abhimanyu
Abhimanyu
Senior Software Engineer
Blake Tickell
Blake Tickell
AI Research Engineer
Sean Spain
Sean Spain
Head of Product/GTM
Brit, Ki, Deloitte

Discover

Understanding the science behind weather forecasting.

Robust Medium‑Term Forecasting Across Extreme Events
Research7 min read

Robust Medium‑Term Forecasting Across Extreme Events

How our physics‑AI models keep signal during unprecedented extremes.

Read more
Today's Extreme Event is Tomorrow's 'Normal'
Technology3 min read

Today's Extreme Event is Tomorrow's 'Normal'

Long-term forecasting under climate change with physics-AI models.

Read more
Forecasting Heat Waves: A Deep Dive
Case Study2 min read

Forecasting Heat Waves: A Deep Dive

Our approach to predicting extreme heat with physics‑informed ML.

Read more

Ready to make catastrophe risk predictable?

Join leading insurers using our forecasts. See ROI in 2–4 weeks.

Email us