PNC Bank is one of the largest diversified financial services institutions in the United States. For thelast three years I have been part of the AI & Intelligent Automation team, a center of excellence thatsupports various business units with their Automation and AI projects.Key Achievements:● Advanced Generative AI Capabilities: Orchestrated the comprehensive assessment andoptimization of RAG evaluation frameworks, identifying key failure points and devising solutions,thereby setting enterprise standards that ensured the successful adoption of generative AItechnologies● Implemented RAG Evaluation Framework: Developed and customized an advanced RAGevaluation framework, employing LLM fine-tuning and advanced statistical methods. Delivered arobust solution optimized for model performance and cost-effectiveness.● Enabled Chatbot Intelligence: Led the implementation of advanced NLP techniques usingKore.ai and Voiceflow, substantially improving NLU and user experience across multiplebusiness units.● Enhance Organizational Knowledge: Spearheaded ad hoc numerous NLP initiatives involvingvector search, topic modeling, keyword extraction, data augmentation, NER, PII detection, Q&Ageneration, and summarization, empowering knowledge workers to make data informeddecisions.● Fostered AI Education: Developed and maintained an NLP-focused knowledge repository,empowering employees with tutorial, how-to guides, explanations and references, designed toenable and build a community of effectual data professionals.● Optimized ATM Cash Management: Addressed critical challenges by developing a customanomaly detection algorithm, a feature engineering library, and advocated for the adoption ofsound data science practices.

Hiiper is an innovative retail analytics platform that provides competitive analysis for categorymanagers. In this concurrent role, I applied cutting-edge AI techniques to solve complex challengesin the retail industry, using NLP, CV, Tabular ML and DL.Key Achievements:● Perfected Online Product Matching: Engineered deep learning NLP models achieving 99.2%accuracy in top-6 results for extreme multi-label product classification and matching.● Empowered Offline Price Monitoring: Utilized object detection, OCR, and reranking to greatlyreduce the amount of human annotations necessary to label offline pricing data.● Enabled Cross Channel Product Matching: Successfully showed you can match on shelf andonline products, by utilizing advanced CV algorithms and active learning, which greatly reducedthe time and cost of adding additional products and/or locations.● Enhanced Data Capture: Incorporated best practices to and advance CV algorithms to improveexisting receipt data and expiration date extraction that saved thousands of hours of humanlabels every month.● Led Research and Development Initiatives: Pioneered advanced research and developmentefforts, creating novel approaches and data cleaning processes that significantly improvedpredictive accuracy and operational efficiency.

Startup enterprise focused on making every aspect of life more effectual. 𝛁 “unbe-LIP-able” App (AppSheet, google sheets, Docparser) • Created a soup-to-nuts multi-platform application that was designed for direct sales marketers. • The CRM portion of the app has the ability for the user to make calls, text messages, email, and social media posts from within the app, as well as activity and opportunity tracking. • Integration with Docparser allows for purchase orders to be emailed by the user and their inventory and product costs are updated in the app to allow for better pricing and avoid orders for products that are out of stock. • Cost-based accounting reports are available on the order, customer, and representative level. • The ability for users to establish customized rewards, promotions and referral programs𝛁 Geo2vec (pandas, Shaply, Geopandas)• This is a multi-year project focused on the creation of geospatial vectors that incorporate, demographic, geographic, psychographic, environmental, and temporal data sources so that the true meaning of locality is taken into consideration. Sub-projects include, address standardization, census data extraction, and much more.

• Write customized SQL queries to retrieve, analyze, and transform data • Use IBM SPSS Modeler to predict customer churn, forecast sales, and perform RFM analysis.• Used R for data manipulation and feature engineering