Greater Toronto Area, Canada
Detail-oriented and analytical professional with 5 years of data science experience using Azure Databricks for big data modelling, predictive/explorative analysis, and machine learning. Involved in end-to-end machine learning, research and management of business opportunities with clients. I currently onboard and train Moneris Data Science teams on Databricks, manage the configuration of the Azure Databricks environment setup (RBAC, Unity Catalogue, Storage Account, etc.) and cost management associated with the cloud infrastructure. Skillset: • PySpark on Azure Databricks using MLlib (pipelines, regression/classification/clustering, streaming, text analysis, model selection and tuning), mlflow, spark streaming • Experience with LLM: fine-tuning, RAG, re-ranking, BERT, MCP, Llama2, T5, BERT, LlamaIndex, HuggingFace • Python (sklearn, numpy, scipy, pd, plt, etc) • Tableau and Plotly/Dash python for visualizations • SQL, Scala/Java, R, Kafka • Some experience with html, css, d3.js, tf, keras, pytorch, xgb, Keras, GPU (Horovod, cuML), VBA, Gitlab, CI/CD, Microsoft Azure
• Involved in projects to monetize data while fulfilling legal requirements to clients such as government agencies, consulting agencies, small and large businesses • Developing an internal facing chatbot (i.e. fine-tuning T5, RAG through BM25/TF-IDF, BERT for reranking) that can answer questions on Moneris corporate documents, engineering standards, anomalies in processing volume and hundred-paged payment terminal manuals • Attended meetings with all large clients to present models and align with their expectations on results and delivery timelines • Built an alert detection model and web application that predicts expected transaction amount in realtime using Spark streaming, producing upper and lower bounds by distributing fb prophet and arima on Databricks, and sends email data if anomaly detected • Analyzed internal cannibalization on chained stores using network algorithm PageRank on merchant location and sales difference over time with new store openings • Created a fraud prediction model using ~60 KRIs on 350,000 merchants resulting in a 450% speed increase in speed of accessing incidents and a 10% decrease in losses from fraud • Developed a distributed geograhical clustering model using dbscan and gmm on billions of transactions that is used by various teams to model residential, working, travel, tourism over time in Canada • Created a network model to predict the cannibalization of sales due to new store openings and optimal locations of new stores based on customer and competitors locations • Attempted to build a speech to text model on audio files to output words and identify different speakers, unfortunately accuracy was not good (ended up using Azure Cognitive Services API)
• Coded VBA translator for EDI files saving $2000 a year in software cost • Researched, contacted, and attended presentations by software suppliers for SEO • Evaluated Samsung website for visitor metrics such as session length, conversion rate, inbound/outbound links, mouse tracking using Adobe Analytics • Analyzed retailer performance using SQL, regression and clustering to determine retailers to target during campaigns • Developed Tableau Dashboards to track KPIs for senior executives and showcased at the Samsung international conference to other global divisions
• Set targets and choose sales promotion programs for all Canadian dealers • Analyzed competitor and industry trends on different vehicle types for executive team • Coordinated national field staff in meeting quarterly and annual targets as well as facilitating communication with internal departments
• Analyzed patterns in commodity index prices to assist traders in market analysis • Entered and modified trades, swaps, hedges into Entero ONE and ICE trading platforms with major CA/US banks and various companies in CA, US and Mexico • Conducted analysis with legal team on trade execution and their impact on performance
• Processed EFT, cheques, employee/company expenses, account clearing and payroll • Utilized Sage 300 ERP (Accpac) and Deltek ERP