All assets referenced here are the intellectual property of Deloitte Consulting LLP ("Deloitte").

Organization

CortexAI for Government - Deloitte Consulting LLP ("Deloitte")

Role

Product Manager (May 2021-January 2023)
Product Development Analyst (September 2019-May 2021)

Overview

CortexAI™ for Government (CfG), Deloitte’s proprietary platform, is an AI/ML blueprint, workshop, and set of solutions for enhancing decision-making and improving mission outcomes for government and public sector agencies.

I've worked in product roles across 2 CfG AI Engines. The first utilizes Natural Language Processing (NLP) to provide analytics on large volumes of text. The second asset uses Optical Character Recognition (OCR) to extract critical information from mass quantities of documents.

Both products began as ideas on a slide deck in January 2021. We have since built each of the products from 0 to 1 and continue to iterate upon them today. Throughout the process, I helped lead an agile transformation, developed/executed product roadmaps, and led product demos for internal stakeholders and prospective clients.

Challenge

Large Volumes of Text

• Case workers have to sift through a large volume of text in order to gain an understanding of what is happening within a specific case.

Mass Quantities of Documents

• Workers have to process mass quantities of documents (IDs, wage documents, etc.) in order to properly administer benefits to citizens.

Solution

Large Volumes of Text -> Case Comment Explorer (NLP)

• Our team built a dashboard that allows case workers to quickly find critical information, get a holistic view of the case, and explore across cases.

• I led the initial user research gathering efforts. After uncovering the need to process cases more efficiently, I then researched 10 different NLP functions that could be applicable. We developed and deployed 4 of these 10 features - summarization, classification, named entity recognition, and sentiment analysis - and packaged them into the dashboard

Mass Quantities of Documents -> Intelligent Document Processing (OCR)

• Our team built a model that extracts critical information from documents, then stores it away in databases for government workers to leverage during the benefits application process.

• AI helps us do this by automatically classifying/sorting the type of document that has been uploaded, accurately identifying important fields, and quickly extracting those fields from the document.

• I led the functional design of the model. After implementing the design in MVP1, I was promoted to Product Manager where I have planned and executed each of the last 5 product roadmaps with a scrum team made up of 5 developers.

Impact

Large Volume of Text -> Case Comment Explorer (NLP) -> Increased Efficiency & Proactive Intervention

• Case workers are able to be caught up to speed much quicker utilizing the dashboard as opposed to sifting through the individual case comments. They can identify themes, keywords, people, places, and time events.

• This allows case workers to identify trends and patterns within a case which could lead to proactive intervention.

Mass Quantities of Documents -> Intelligent Document Processing (OCR) -> Faster Execution in Government

• Workers are able to process applications for benefits much quicker utilizing the extraction model as opposed to manually reviewing individual documents. They can verify information and uncover, escalate, and eliminate common barriers during application review.

• This speeds up the application process and ultimately ensures faster distribution of benefits to citizens.