Looking for a job

Yi Huang

@yihuang

Senior data scientist at Constellation Energy forecasting river conditions and developing audit models for energy operations.

United States

Message

At Constellation Energy, I forecast intra-day river conditions, reducing forecast error by 10% to inform renewable asset operations. I also developed an audit model that identifies discrepancies across four document types and internal systems of record with 97% accuracy.

At Bayer, I used causal inference and time-series models to inform decisions across product and commodity markets, and led data science projects from research questions through stakeholder adoption. My Ph.D. research at MIT produced 16 publications and more than 1,000 Google Scholar citations on numerical modeling of wearable materials, solar cells, and optical coatings.

Experience

Work history, roles, and key accomplishments

Current

Senior Data Scientist

Current

Constellation Energy

Jul 2024 - Present (2 years 4 months)

Forecasted intra-day river conditions to reduce forecast error by 10% and developed an audit model for natural gas deal flow with 97% accuracy. Contributed to battery auction algorithm using Markov decision process models and partnered with engineers and stakeholders to drive model adoption.

Python R SQL AWS GCP BigQuery MLOps Markov Decision Processes Time series Forecasting Data Pipelines

Data Scientist

Bayer

Aug 2021 - Jun 2024 (2 years 10 months)

Used causal inference to measure external factor impact on sales across 50+ products and developed time-series models to forecast commodity prices, reducing prediction errors by over 10%. Led data science projects end-to-end and mentored junior data scientists.

Causal Inference Time series Forecasting ETL Python R SQL AWS GCP BigQuery Stakeholder Management

Data Scientist

Standard Chartered

Feb 2021 - Jun 2021 (4 months)

Developed time series forecasting models for macro-economic indicators, outperforming state-of-the-art Fed dynamic factor models by 10%. Constructed data pipelines from GB-size financial data sources to evaluate fundamental attractiveness of EM corporate bonds.

Time series Forecasting Data Pipelines Python R SQL AWS GCP BigQuery