Company

Turning the weather into an advantage

We want every renewable asset dispatched with foresight. That means forecasts that know each site, decisions that hold up across the range, and a clear record of both.

MissionMission

Every asset, dispatched with foresight

Forecast
asset-level, calibrated
Decide
explained, within limits
Learn
from every settlement
Example. Values are illustrative.

Why we started

Clean power should not be punished for being variable

Wind and solar operators pay every day for weather they could not see coming: imbalance penalties, curtailed output and trades made by hand under time pressure.

AI weather models and physics-ML have changed what a forecast can be. We started Helivane to carry that all the way to the site and on to the decision.

Principles

How we work

Measure, then claim

Results come from pilots

We publish numbers only after we have measured them on real sites.

Uncertainty is information

Show the range

A forecast without a band hides the thing a desk most needs to know.

People set the limits

Autonomy is earned

Each step toward automation is a choice our customers make.

Roadmap

What we are building, in order

Draft roadmap. Dates depend on pilots and may change.

StageFocusStatus
NowSkycast downscaling and calibration on pilot sites; imbalance attributionIn progress
NextDispatra at L0–L1 for intraday; Cellix for storage pilotsPlanned
LaterFleetra portfolio optimization; Twinora edge pilots; Tempora librariesPlanned
BeyondL2–L3 autonomy, proprietary global weather model, more marketsLater

Careers

Open roles

Draft list. Write to careers@helivane.com with the role in the subject line.

Founding ML engineer, weather and physics-ML

AI weather ensembles, downscaling and Modulus surrogates.

Hybrid, EU

Optimization engineer

Stochastic optimization and cuOpt for bidding and portfolio planning.

Hybrid, EU

Energy markets lead

Day-ahead, intraday and balancing market design and pilot delivery.

EU

Full-stack engineer, operator tools

The desk-facing product: forecasts, decisions and audit trails.

Remote, EU time zones

Edge and OT engineer

Jetson, Holoscan and safe integration with site control systems.

Hybrid, EU

Stack

What we build with

NVIDIA Earth-2Ensemble AI weather and km-scale downscaling
NVIDIA ModulusPhysics-ML surrogates for terrain, wakes, irradiance and cells
RAPIDSGPU processing of SCADA and geospatial time series
cuOptStochastic scheduling and fleet optimization
NeMo and GuardrailsThe dispatch agent, its rationales and its limits
Triton, TensorRT and NIMLow-latency serving across many assets
OmniverseThe portfolio digital twin
Jetson and HoloscanEdge inference and sensor streams at the site
CosmosSynthetic extreme-weather scenarios
Python and CUDAModels, optimization and data pipelines
TypeScriptOperator and trader tools

Build the forecasting layer for clean power

If you work on weather, optimization or energy markets, we would like to hear from you.