Work Experience

• May 2022 - present: Senior Data Scientist, Rogers Communications
  • Spearheaded the end-to-end development and deployment of J.A.R.V.I.S, a Retrieval Augmented Generation (RAG) pipeline using Databricks with Llama 3.2 as the LLM model. Enabled non-technical teams to query 100M+ network KPI records via natural language, reducing manual data retrieval time by 30%.
  • Created a forecasting model that leverages Meta’s Prophet to predict network traffic growth. Takes into consideration seasonality and the influence of public events on abnormal network traffic.
  • Developed a real-time network disruption identification tool. Collaborated with network engineers to identify problematic network patterns and trained a Support Vector Machine (SVM) model to identify these patterns in live data. Enabled preemptive issue resolution and reduced response time from hours to minutes, ensuring uninterupted service to millions of customers.
  • Leveraged crowd-sourced customer usage data, network KPIs, and geospatial data to train an ML-based (Random Forest) tower coverage model. Utilized by national planning to evaluate locations for new towers. Automated and optimized many network planning aspects, reducing Capex by over $400 million.
  • Spearheaded the development of a forecasting model leveraging Meta’s Prophet to predict network traffic growth, integrating considerations for seasonality and the influence of public events on abnormal network traffic.
  • Designed a comprehensive and intuitive KPI dashboard utilizing Grafana to monitor the ongoing Rogers 5G deployment in the TTC (Toronto Transit) subway system. Used extensively by senior leadership, planning teams, and external stakeholders, this dashboard was pivotal in the deployment of cellphone coverage in less than 4 months.

  • Utilized customer connectivity data to provide customer journey analytics, station congestion, and other dashboards and analytics to TTC. This was part of a pioneering initiative to leverage customer connectivity data to collaborate with municipalities and improve municipal services planning.

  • Led cross-functional teams: Successfully led and coordinated efforts between data science, network engineering, and executive teams to implement innovative solutions.
  • Strategic Planning: Played a key role in the strategic planning and execution of network infrastructure projects, significantly enhancing operational efficiency.
• June 2021 - April 2022: Data Scientist, Rogers Communications
  • Led a 911 criticality study using crowd-sourced data and network KPIs to identify critical sites whose coverage areas have no backup from any network operators. Used in assigning criticality to sites to ensure no area loses 911 connectivity.
  • Determined appropriate techniques, models, and extensions to analyze geospatial data using GeoPandas, Shapely, PyKML, MongoDB, PostGIS, and various SQL functions.
  • Developed a tool that utilized a Random Forest Classifier on network quality data to detect and flag degradation in customer performance in real time, reducing response times by over 78%.
  • Proactively refactored and enhanced the performance of ETL pipelines through the implementation of intelligent partitioning and indexing. Achieved a 64% reduction in run time.
  • Integrated terabytes of crowd-sourced customer and geospatial data from multiple sources at different h3 hexagon resolutions. This data was then visualized in an interactive web interface, allowing planners to drill down to a 65 m resolution.
  • Provided mentorship to new team members through pairing sessions, offering an understanding of the internal data stacks and methodologies.
  • Led a team of 3 to migrate multiple on-prem processes onto the Azure cloud environment.
• June 2020 - June 2021: Software Developer, Rogers Communications
  • Provided metrics for API usage for stakeholders and performed ad-hoc analysis for the app ecosystem using SQL, Jupyter Notebooks (Pandas, SciKit Learn, and Facebook’s AI libraries) to support ongoing projects in the partner-facing product areas.
  • Developed an automation tool for a nationwide audit of network parameters. Utilized by all regional teams, reducing OPEX costs by $465 thousand/year and avoiding a Capex of $1.3 million.
  • Conducted feature analysis on customer data to identify key causes of churn. Developed a random forest classifier to classify high churn-risk customers for retention marketing.
  • Analyzed large sets of data (Python, Pandas, SQLAlchemy, PyMango, and MariaDB), developed machine learning models (Python, PySpark, AzureML, MLflow), and created complex dashboards (PowerBI and Grafana).
• May 2018 - May 2019: Software Development Engineer Intern, AMD.
  • Developed a comprehensive testing and verification framework enabling engineers to validate functionality pre-integration, ensuring seamless deployment to the main chip.
  • Worked on synthesis, timing closure, and formal verification on the block and/or chip level using state-of-the-art CAD tools.