Databricks Data Lakehouse Architect – Canadian Public Sector Developed a data ingestion framework to load 100+ tables from Oracle on-prem to ADLS Gen2 in ADF and designed a Lake house utilizing Databricks Medallion architecture to create 5 data models that facilitated the re-engineering of 10+ Cognos reports to PowerBI (Azure Data Factory, Databricks, PowerBI)Azure Data Engineer – Tier 1 Insurance Company​Decommissioned a no/low code ETL tool to a custom Metadata ingestion framework utilizing Azure Data Factory, Databricks and SharePoint to enable Data Scientists/Engineers to migrate 10+ applications, 30+ data sources from on-prem to Azure Data Lake Storage 2 and SQL Server Managed Instance (Databricks, ADF, Azure SQL MI)Azure MLOps Engineer – Top 3 Canadian Telecom Company- Built an MLOps framework to deploy ML models to production using Azure DevOps and Azure ML.- Developed a low-latency real-time inference pipeline that predicts customer intents from chat data using delta live tables in Databricks (Azure DevOps, AzureML, Databricks)AWS ML Architect – Top 3 US Telecom CompanyLed the architecture design of the service ticket engine that leverages 12+ ETL pipelines and 3 machine learning models (Proactive chronic node detection, Edge health score forecast, and Probability of network service affecting events) to submit proactive maintenance tickets to reduce over $6 million dollars of yearly transactional cost. (AWS Sagemaker, Lambda, Neptune, Jenkins, Step functions)AWS MLOps Engineer - Top Canadian Mining CompanyBuilt an MLOps framework that automated data pre-processing, model training, batch inference and model deployment using AWS Sagemaker Studio, S3, Lambda and Step functions.

AWS Data Engineer - Top-5 Canadian Bank- Leveraged Amazon Connect to build contact flows for a virtual call center and integrated AI services like Lex, Comprehend and Transcribe for chatbot capability, speech-to-text transcription and sentiment analysis (AWS Connect, S3, Lambda, DynamoDB, Comprehend, Transcribe, Lex)AWS Data Engineer - Tier-1 Canadian Insurance Company- Built a data processing pipeline to transcribe audio calls of agent customer interactions and indexed the resulting call transcription & metadata in Elasticsearch. (Python, AWS Lambda, Transcribe, S3, Fargate, Step functions, Elasticsearch)

Mileage Optimization for a Rental Car Company- Built Machine Learning models to classify fleet and customer population so that the right car is paired to the right customer by load balancing mileage across vehicles.- The fleet and reservation indicators were deployed in AWS in both real-time and batch process using Sagemaker, Postgres, Lambda and DynamoDB.Recommendation Engine for a Series E funded startup- Built a recommendation engine framework leveraging the WALS (Weighted Alternate Least Squares) algorithm in Tensorflow to predict future purchases by end users.- Deployed the WALS model on a Kubernetes cluster using Docker, AWS S3 & EC2.(Tensorflow)Cyber Insurance NLP POC for a Tier-1 Insurance Company - Used natural language processing to automatically extract key policy information from PDF documents thus reducing manual effort from underwriters. - Created a competitive advantage with both IP and enhanced NLP capability through a proprietary lexicon of cyber insurance coverage terms. - Reduced overall risk and exposure by incorporating extracted information into risk models.(Python - SpaCy, NLTK, Gensim)Deployment of Sales Forecasting models in Azure for Top-3 CPG Company- Developed three predictive models using ensemble methods to forecast at a SKU level eCommerce sales, orders, and possible disruptions.- Built pipelines in Azure Data Factory for data preprocessing/feature engineering and deployed models in R (caret) using HDInsights.(Scala, R, Azure Data Factory, Azure HDInsights)Digital Modernization of Operational Data Store (ODS) for a Tier-1 Financial Services Company in AWS - Wrote ETL Scripts in Apache Spark (Pyspark) in Zeppelin notebooks to calculate account balances and asset performance of participants.(Pyspark, Zeppelin, AWS S3, Glue, Athena)

• Cleaned, structured and merged 15+ providers related to credit bureaus, digital footprint, property valuation and location data for 75,000 small business prospects in SQL and Python for credit risk rating proxy model.• Built a company revenue growth model utilizing Gradient Boosted Trees in Python and DataRobot to identify businesses in top decile that are 17x more likely to grow in the next year at 10% or more than the sample average.• Built a pre-screening credit model leveraging Equifax data using Logistic Regression with one-hot encoding in R, which buckets customers in deciles ranging from 3% to 22% delinquency.• Developed a KYC (Know Your Customer) solution from 10,000 auto loan applicants employing waterfall attributes in Python through multiple external data sources, resulting in 40% automation of manual verification checks.• Configured attributes in Python to access risky behavior of 150,000 applicants for life insurance underwriting using Bing & Twitter API, sanctions & watch lists and criminal data.• Built Machine Learning models including Logistic Regression, Naive-Bayes, SVM, Random Forest, Gradient Boosted Trees, XGboost, RuleFit Classifier and GLM's to optimize client KPI's and visualize output results in the form of ROC curves, Lift charts, Partial dependence plots and Reason code analysis.My Technology stack includes R (ggplot2) for visualization, SQL (PostgreSQL) & Python (Pandas) for data wrangling/manipulation and DataRobot/Scikit-learn for Machine Learning.

• Assisted and graded programming assignments of 60 students in Python for Web Analytics (BIA 660) course.

• Estimated an additional $200,000 in revenue by applying Gradient Boosting algorithm (GBM) in R to predict response rate of direct mail marketing campaign.• Reduced report development time by 91% using T-SQL and Tableau by automating generation of market share reports from FDIC website.• Prepared data for credit risk modeling by extracting, merging and sampling from more than 1 million records in multiple credit bureau (Experian) tables utilizing T-SQL and Excel. • Implemented binned Logistic Regression model in R to predict probability of future applicants defaulting on auto loans.• Designed interactive risk management dashboards in Tableau for unsecured personal loans portfolio.• Minimized customer service costs by identifying 5 branch locations for ATM deployment, utilizing advanced data exploration in Tableau.