As an Applied AI Consultant, my role is to provide clients with expert advice on integrating AI technologies intotheir business operations. I work closely with clients to understand their specific needs and challenges, and developcustomized AI solutions to meet those needs. This includes designing and implementing AI algorithms, models, andtools, conducting feasibility studies, and delivering training programs. My goal is to help clients improve theirbusiness processes and decision-making using the latest AI technologies and methodologies.I have worked on the following projects:● Developed an architecture blueprint for a Fortune 5 healthcare company to implement Machine LearningOperations in the Google Cloud, Vertex AI environment. Developed this blueprint for multiple use-case such asTabular, Image, Text, Video and Audio. In addition developed a pod structure and governance to implement thiswithin the company.● Developed a Natural Language model to recognize the intent of callers calling into the call center of a Fortune 5financial institution in order to accurately route customers to the correct call queue inorder to reduce the transferrate in the call center.In addition to my client work I work on various firm initiatives:● Table to Text Summarization using GPT-3● Using the deep learning Transformers architecture for Time Series problems.● Semantic Search Engine for quick document search

Working on expanding the Voice of Customer tool to an Enterprise wide product.

Currently Working in the Conversational A.I team to leverage speech and text data in the contact center and digital feedback space using advanced Natural Language Processing. Using completely open source technologies to build an end-to-end Voice of Customer tool to be leveraged by both executives and analysts. - Data Ingestion from third party vendors using REST API’s.- Data pre-processing, cleaning and transformation using, pypark, pandas, nltk, and spacy.- Applying large open-source language models such as BERT to classify digital feedback.- Making use of Spotify and Google technologies to build a real-time customized aggregated search engine.- Bayesian methods to identify upcoming trends in customer complaints and feedback in real-time.- Interactive front-end display in Shiny, ability for users to directly interact with complex models.- High level KPI tracking for executive view.- Complete product deployment in AWSI am also responsible for developing and maintaining Region’s in-house language modeling Python package, for use in various other projects in the bank.

• Given a product database the task was to determine which plants were producing the highest percentage of faulty products and then determine the reason behind the faulty production. In addition, the team was to create a predictive model in order to determine a faulty product before it left the factory. • Given the sales database the task was to determine which segment of dealers was most beneficial to Hunter-Douglas and identify the core reasons for them being so.

Afiniti provides real-time optimized pairing of individual contact center agents with individual customers in large enterprises by applying artificial intelligence to discover, predict and affect patterns of interpersonal behavior.Worked as part of the AI-Production team as an AI manager for the DGS account and was part of the T-Mobile and SKY AI team, the company’s two largest accounts. Daily duties included:● Analyzing new data for any change in trend or customer behavior and then modified agent-caller assignment algorithm accordingly. ● Making sure that the performance dashboards were accurate and up-to date so that senior management could have clear visibility of how an account was performing. ● Daily meetings with both internal and client-side account managers to stay up to speed on performance-reporting and ensure poor performance was mitigated and improved.How we provided benefit to our clients:● Increased T-Mobile’s revenue up to $6M per month. ● Provided an annual increase of around 50,000 customers and reduced churn rate by approximately 6% for SKY UK. ● Worked closely with the R&D team to test our latest algorithms. Used a Linear Program (Network-Flow) to provide intelligent agent assignments to callers at run-time; a Hierarchical-Bayesian Model was used to evaluate the cost of each agent-caller interaction, which was calculated using the STAN probabilistic programming language and Bayesian inference engine, through a Hamiltonian Monte Carlo approach to approximate posteriors. ● Oversaw the testing of this new modeling approach in production and created a streamlined approach to creating, deploying and monitoring the model.