OVERVIEW
Procter & Gamble (P&G) is a multinational corporation specializing in a wide range of personal health/consumer health, and personal care and hygiene products.
During my co-op in Boston, I was a part of P&G's grooming function, working with prominent brands such as Gillette, Braun, Venus, and King C Gillette. Through my projects, I collaborated with international teams from the UK, Brazil, Mexico, Poland, and China.
👩💻 Type: Co-op
💡 Role: Data Scientist
🗓️ Timeline: June 2023 - Dec 2023
PROJECTS
This section aims to give a brief overview of my projects. This section is intentionally vague to prevent disclosure of sensitive information, focusing on the high-level challenges and strategic considerations of the project.
Project 1: Asset Value Optimization
How can we maximize the ongoing value of the equipment in the plants?
This project centered on equipment obsolescence. Equipment obsolescence is when an equipment is no longer optimal for use, either because of wear and tear, or evolving operational needs.
Problem
Annually, team members convene to evaluate the plant’s equipment inventory, identifying pieces that may be considered for obsolescence. This process involves critical decision-making on whether to decommission, transfer, upgrade, or continue using the equipment. A key aspect of these decisions is the determination of whether to dispose of equipment, especially considering industry norms where acquiring new primary machinery could exceed a year. The project tackles three challenges which lead to costs and resources leakages.
1. There's a trend where decision-making in this domain tends to rely more on individual expertise and intuition rather than being heavily data-driven. This approach means that much of the critical knowledge is person-dependent rather than being systematically integrated into organizational processes.
2. There's a lack of a unified, global system for tracking and managing the lifecycle of packaging equipment. This results in varied processes across different locations, leading to inconsistencies and potential inefficiencies in handling equipment obsolescence.
3. The project identifies a general challenge in visibility regarding the future obsolescence of equipment. This limitation potentially impacts the ability to strategically innovate and prepare for future needs in equipment technology and capabilities.
Project Contributions
In this project, I closely led and worked with a team of data analysts, vendors and engineers. I reported our findings to Senior Director of Process & Engineering on a weekly pages.
First, I began the project doing a stakeholder analysis and placing every group in a power-interest matrix. This allowed me to have a holistic understanding of key players involved and help set the impact/vision of the project. I dove into meetings and interviews with these groups to have an in-depth understanding of their workflows, pain points, and priorities.
Second, I designed an Obsolescence Review Workflow that contained an evaluation criteria for obsolescence based on certain data points, approval/escalation processes, and time stamps from start-to-end. This workflow went through multiple iterations through feedback sessions with numerous stakeholders.
Third, we began built a pilot dashboard for a set of plants. This dashboard adapts the workflow we created to automate obsolescence evaluation process.
I presented my findings and progress to a over 200 employees in R&D. Multiple other groups demonstrated strong interest in adopting our Obsolescence Workflow and tool for their own categories of equipment. This would introduce thousands of additional equipment line, multiplying the current impact of $6MM of savings. I pitched to the VP of Process of Engineering (highest authority in Boston office) with 4 Senior Directors, who were impressed by the work and decided to invest more resources into the project for expansion.
Project 2: Image Processing
How can we quantify the entire shaving experience and extract valuable data from images pre and post shave?
This project's goal was to support emerging skincare studies, focusing on quantifying the shaving experience through image analysis. This approach was different from current traditional qualitative consumer testing methods. We used image data to derive insights about post-shave factors such as closeness, skin smoothness, and hair count. This opens up new avenues for innovation, especially in validating consumer perceptions against the actual performance of razors on the skin, and in re-evaluating product claims.
Project Contributions
First, I developed a comprehensive data management pipeline for handling image data. This involved gathering images from multiple sources, extracting essential information, and organizing this data in tabular format using Apache Spark and Databricks. This systematic approach was crucial for ensuring the accessibility and consistency of the data, thereby facilitating more effective and accurate analysis.
Then, I ventured into Image Quality Assessment where I created an algorithm to evaluate image quality. This algorithm incorporated various models categorized into reference-less (e.g., BRISQUE, NIQE) and reference-based methods. Using these models, I established a benchmarking process to assess the quality of images, identifying and filtering out low-quality images from our datasets. This was vital as the quality of the images directly impacted the reliability and accuracy of our data analysis.
Project 3: Packaging Intelligence Dashboard
How can we assess current capabilities and build new capabilities that add value to the existing dashboard?
This dashboard is designed to identify market gaps, innovation opportunities, and potential cost savings. It serves as an aid for partners and development teams. It analyzes the packaging portfolio across various dimensions such as product lines, packaging platforms, markets, and over time
Project Contributions
My contribution in this project were multifaceted:User Engagement and Analysis: I conducted interviews and surveys with dashboard users, coupled with an analysis of past data and user experience metrics. This process was essential for identifying improvement areas.
Recommendation and Implementation: Based on my findings, I developed recommendations to optimize the dashboard. I then implemented these critical features.
Talent Analysis and Upskilling Initiatives: Simultaneously, I performed an in-depth talent analysis within Gillette R&D. Discovering a knowledge gap and external dependency on third-party contractors, I created training modules to drive digital upskilling. These efforts culminated in delivering strategic insights to the VP of R&D and developing training materials focused on digitization.