The Data Guild helped develop and assess models and key performance indicators (KPIs) for energy usage. Designed methods to measure change after some event (causal inference) using counter-factuals, and corresponding uncertainty. Provided strategic direction and best practices related to code, machine learning systems, and parallelizing analysis.
- Benchmarked against the ASHRAE 14 definition of savings
- New definition of energy savings based on counterfactuals & building models
- Machine learning (ML) pipeline parallelization
- Technologies utilized: python and scikit tools, starcluster spot instances on AWS compute.
Aman designed the ML pipeline and parallelization, implementing the savings calculation at scale. I worked with the team to assess our new savings calculation against the benchmark standard.