World-class technology

Electricity Maps; United States; Remote; USD 80,000 - 140,000; Full time

Job summary

Join a forward-thinking company that leverages cutting-edge technology to deliver real-time power flow analysis and forecasting. As part of a dynamic team, you will work with advanced machine learning algorithms to process vast amounts of electricity grid data. Your role will involve developing models that ensure data accuracy and performance, while also contributing to innovative solutions for anomaly detection and forecasting. This is an exciting opportunity to make a significant impact in the energy sector, using your skills to enhance the reliability and efficiency of electricity grids worldwide.

Qualifications

• Experience with machine learning algorithms and data processing.
• Strong background in data validation and anomaly detection.

Responsibilities

• Develop and implement machine learning models for forecasting.
• Ensure data accuracy through validation and correction processes.

Skills

Machine Learning

Data Analysis

Forecasting

Anomaly Detection

Data Validation

Education

Bachelor's in Data Science

Master's in Computer Science

Tools

Python

TensorFlow

SQL

Job description

Enterprise-grade proprietary data platform performing
real-time power flow analysis and forecasting at a global scale.

The Electricity Maps platform processes and aggregates live electricity grid data, including energy production, consumption, and electricity flows. It leverages advanced machine learning algorithms and extensive integrations with grid operators worldwide to provide highly accurate calculations. The platform dynamically models electricity flows using the flow-tracing methodology.

Ingesting billions of datapoints daily

The platform integrates with over 100 grid operators worldwide and collaborates with market operators and weather providers to power its forecasting engine. This results in the ingestion of billions of datapoints every day.

Highlights

• Integrates with 100+ grids worldwide, covering more than 350 zones for extensive geographical coverage.
• Ensures data is always up-to-date through automatic refetching after days, weeks, and months, consolidating data for availability.

Detecting & correcting anomalies automatically

Data from grid operators can sometimes be delayed, missing, or incorrect. Specialized anomaly detection algorithms identify incorrect data, which is then updated using machine learning-based synthetic data models, maintaining high data accuracy and availability.

Continuous automatic data validation removes invalid data and replaces it with synthetic estimates.

Synthetic estimation models

Multiple machine learning-based models generate synthetic data during data delays, gaps, or invalid entries.

Forecasting Engine

The platform forecasts the grid's state, including electricity flows, day-ahead prices, electricity mix, and carbon intensity, up to 72 hours ahead. This is achieved through regression and tree-based methods, which are regularly retrained and backtested to ensure optimal performance.

Forecast models are continuously retrained, backtested, and monitored in real-time.

Performance Reports

Performance metrics are available via a dedicated dashboard for customers.

Flow Tracing

To attribute emissions accurately, the platform traces the origin of electricity, accounting for transmission line effects. This ensures emissions are linked to the actual power plants, not just the consumption location. On average, over 10% of emissions are from imported electricity, with errors ranging from 10-15% up to 90%.

Since 2016, the flow-tracing methodology has been peer-reviewed and is recognized globally.