Open to opportunities

Michael Lee

@michaellee6

Staff software engineer specializing in large-scale recommendation systems and ML infrastructure, driving measurable personalization improvements.

United States

Message

What I'm looking for

I’m looking for a team where I can lead end-to-end ranking/personalization and ML platform work—designing real-time pipelines, owning reliability, and partnering cross-functionally to ship experiments that measurably improve engagement and retention.

I’m a Staff Software Engineer with 10+ years building large-scale recommendation systems and ML infrastructure. At Yahoo News, I help power the ranking and personalization stack serving tens of millions of monthly active users, from candidate retrieval through real-time inference.

I led the design and evolution of a multi-stage recommendation pipeline using Python and Java/Scala on Apache Spark. I architected a real-time feature store (Kafka, Flink, Redis) to cut feature freshness latency to under 5 seconds, redesigned candidate retrieval with FAISS to increase long-tail recall by 34%, and migrated to a two-tower deep learning-to-rank model that improved click-through rate by 11% and session depth by 7% in production A/B tests.

I also drive platform strategy and execution: migrating ranking/serving infrastructure from GCP to Kubernetes on AWS (EKS) with Terraform and Helm, reducing new model deployment time from days to hours and compute costs by ~20%. As a technical mentor and on-call owner, I led incident response and postmortems to raise service SLA from 99.9% to 99.97%, and built evaluation/experimentation frameworks that cut validation time from two weeks to three days.

Experience

Work history, roles, and key accomplishments

Current

Staff Software Engineer

Current

Yahoo

Jan 2024 - Present (2 years 6 months)

Led the design and evolution of Yahoo News content recommendation and personalization systems, including a multi-stage pipeline powering personalized feeds for tens of millions of monthly active users. Improved feature freshness latency to under 5 seconds, increased long-tail recall by 34%, and migrated ranking/serving infrastructure from GCP to an AWS EKS platform.

Senior Software Engineer, ML Platform

Yahoo

Jan 2021 - Jan 2024 (3 years)

Built and operated the online model-serving and ML platform capabilities for Yahoo News ranking and personalization models. Productionized user embedding and experimental ranking pipelines, including TensorFlow Serving deployment patterns and real-time experimentation on Kafka event streams.

GCP GKE Istio Traffic Splitting Automated Rollback Spark Python Implicit Feedback Embeddings Java Scala scikit learn TensorFlow Kafka A B Experimentation Python Statistics Engine Scipy React Redux Data Pipelines

Software Engineer, Search & Personal

Yahoo

Jan 2018 - Jan 2021 (3 years)

Developed backend services for personalized search result re-ranking and integrated user history signals into the ranking pipeline. Built and migrated large-scale event logging/feature extraction systems and helped move legacy batch scoring toward near-real-time scoring.

Java Ranking Kafka Hadoop Hive MapReduce Feature Extraction JavaScript React Spring Framework

Full Stack Developer

Verizon Communications Inc

Jan 2017 - Jan 2018 (1 year)

Built frontend UI/UX enhancements and integrated them with backend services for My Verizon and enterprise dashboards. Improved performance by optimizing database queries and implementing Redis/Memcached caching, reducing page load times by 40%.

Frontend Backend Services Caching Strategies Redis MemCached