Senior Data Scientist
CurrentI create, deploy, and monitor models that are consumed by other RealPage departments and external vendors.I work on and actively develop in R and python: - multichannel attribution models using historic performance, sampling techniques and markov attribution models for multiple verticals we serve - lead models using NLP on two-channel call transcriptions - lead models using ensemble techniques that consume sequential and tabular data in tensorflow and lightgbm - attribution features that directly feed multichannel attribution and lead models - geospatial models that leverage clustering for ad targeting using historic migration patterns - ML Ops using codefresh for ci/cd, dataproc with Airflow 1 and 2, BigQuery and Buckets for data consumption and model output storage, and multiple internal and external vendors to assist with data ingestion and ensuring data integrityI am also a model and data logging evangelist who drove the adoption of CometML at RealPage. I work with Wallaroo.ai to provide shadow deployments with challenger models that compete against prod models while offering last mile model deployment layered on top of GCP's Vertex AI to slash API inferencing costs by 80% of GCP's offerings. I also support other departments by offering a/b testing for attribution strategies and new vender offerings (e.g. smart bidding strategies vs manual bidding).
