• Deploy email classification model to enable the front office team to automatically categorize and assign the case to the requestor
• Increased customer satisfaction due to frequent and fast resolution and cost savings as bots are automated solutions
• The solution contains two modules vectorization module has models like Word2Vec, GloveVec, Custom vocabulary and the classification model has conv1d followed by neural network layers
• It is deployed in Azure virtual machine through API it is communicated with salesforce

• Detecting beer brands on Visi cooler screenshots shared by vendors of ABInBev to validate the utilization capacity of ABI brands
• Quantify the allocation of beer brand in VISI coolers to monitor the usage from vendors
• OpenCV helps in data preparation in cleaning, sizing and format the image data in the binary format along with coordinates

• Improve OTC collections and cash position by predicting delay in payments to enable the customized collection actions
• Design develop and deployed a full-stack intelligent led collection tool using python and deploy it on local Mainframe for the Canada region
• Taking pre-emptive actions can drive down the collection time by 1 week
• Optimize the use of collections resources by prioritizing the delinquent invoices for follow-ups
• The champion model selection has Logistic regression, Gaussian Naive, Decision Trees, Random Forest, KNN,XgBoost, Linear Discriminant Analysis

Developed a Resume Scoring Tool for the Talent Acquisition team leveraging historical data to intelligentlyrank profiles based on selected candidates for specific job codes.The tool also extracts key information such as personal details skills and experience storing it in astructured database.This solution streamlined the candidate selection process improved hiring efficiency and enhanced thequality of hires by ensuring better alignment with job requirementsKey skills used in this project included NLP NLTK SpaCy word embeddings TF-IDF PySpark and AWS.

Designed and implemented a Bankruptcy Prediction model for the Order to Cash team leveraging machinelearning techniques to forecast potential bankruptcies among clients.Improved risk management and reduced financial discrepancies achieving a 10% reduction in the bad debtratio.Collaborated with finance and operations teams to integrate the model into existing workflows enhancingoverall efficiency and decision-making.Key skills used in this project included Pandas scikit learnXGBoost and MatplotLib

Developed a data-driven Cash Flow Forecasting system using ARIMA LSTM and other time series modelsto provide more accurate forecasts of account receivables and payables for the treasury team.This system enhances cash flow management by enabling precise financial projections supporting strategicdecision-making and resource allocation.Additionally it facilitates accurate funding requests for operational expenses related to cash payables andreceivables from banks ultimately increasing cash flow efficiency.

Spearheaded the development of an Engineered a solution that converts scanned medical documents intosearchable formats enhancing accessibility and enabling efficient information retrieval.Collaborated with medical experts to design and implement advanced algorithms for accurate dataextraction and document conversion.Improved operational efficiency and accuracy by automating tedious manual tasks related to claimprocessing and document management within the healthcare domain.Key Skills used in this project are OpenCV LlamIndex Langchain OpeAI ChromeDB GenAI LLM Vectordatabase