Joseph Lane Email & Phone Number

San Francisco, California, Us

Contributed to the Support Genius team to develop a company wide LLM to support all Pinterest infrastructure including SQL and database support using Langchain and vector embeddings.Utilized Python, R, SparkSQL, PrestoSQL, Tableau, Dash, and Superset to assist the PMM and PMI team with organization-wide analyses on quarterly performance and projections. Additionally developed sales and ad KPI's for Multi-Objective marketing among all verticals and countries. Created an auto-tagging system harnessing the capabilities of SetFit, an efficient framework for few-shot fine-tuning of Sentence Transformers, to enhance the accuracy of ticket categorization within Zendesk support. The primary objective was to replace a third-party tagging solution, presenting a potential cost-savings of approximately $1.2 million.Pioneered and taught Measurement Research the application of meta-analysis techniques in brand lift studies. This approach unearthed critical insights into the relationship between various Multi-Objective factors and brand lift effects, helping the Product Marketing Management team further sell MO.Developed a Python library to facilitate the creation and upkeep of Measurement Research analytic tools, leveraging CI/CD best practices. Internally deployed and distributed the library using Jenkins, Artifactory, and Github. Improved Multi-Objective study testing for our team of analysts by creating customized tools. Applied statistical best practices to prevent issues such as P-value manipulation, confounding variables, inadequate sample sizes, and improper hypothesis testing selection.Created performance metrics and reports about Pinterest's largest global clients and presented findings to global sales management.

• Constructed new metrics and reports to measure the quality of relationship among clients using RedShift, PostgreSQL, Python and Jaspersoft Server. Ultimately, helping the sales team maintain each client's loan pipeline, PMP underwriting/closing performance, and overall client satisfaction.

• Managing and maintaining qopbaseball.com through HTML, CSS, JavaScript, R, and SQLite on a Linux server. • Seasonal data is analyzed and made available on QOP calculator.

New York, Ny, Us

Market analysis on all Clorox subsidiaries through Traditional Television, Internet Marketing, Affiliate Marketing, and many other campaign types.

Livermore, Ca, Us

Contract Position based on project. Once project was completed, employment ended.• Established initial investigative methods for Ground Penetrating Radar (GPR) research, saving over 3 months of research through a Machine Learning pipeline.• Reduced time needed for exploratory analysis by several months through verifying best radar performance among 36 different radars types.• Built a machine learning pipeline in Python for comparing the performance of GPR technologies, used for bomb detection. This included PCA reduction, logistic regression, model storage, k-means clustering, precision-recall curves, AUC-ROC's, Andrew's curves, and confusion matrices.• References and Letter of Recommendation available upon request.

La Mirada, Ca, Us

• Helped double analytics team membership and train unexperienced members to become full contributors.• Team analyst on a variety of MLB controversies through a patented statistic (Quality of Pitch—QOP™).• Team lead in the investigation on the 2017 and 2019 home run controversies, quantifying pitch deception, predicting injury susceptibility in pitchers, and illegal sign stealing detection. Resulting work and papers produced 4 publications by the world's most famous MLB publishers.• Data wrangling, visualizing, and analysis through R (ggplot2, plotly, dplyr, tidyr,...etc.), Python (matplotlib, sklearn, keras...etc), and Excel. Various uses of GLM's, Neural Networks, and clustering.

Master Of Science - Ms, Computer Science

Bs, Applied Mathematics/Data Science

Bs, Western Classics