Jekyll2026-02-11T08:37:59+00:00https://sagrd.github.io/feed.xmlSagar DhungelSagar Dhungel - A Data Science and Engineering Consultant. Sagar DhungelLarge Scale Health Data Monitering With Databricks2025-12-21T00:00:00+00:002025-12-21T00:00:00+00:00https://sagrd.github.io/large-scale-health-data-monitering-with-databricksA wearable healthcare company was generating massive high-frequency data in Azure Blob Storage, tracking metrics like heart rate, respiration, oxygen levels, and activity every minute.

healthband er diagram

Scope:

  • Instantly alert users when heart rate crossed thresholds.
  • Send personalized notifications and celebrate user milestones.
  • Warn users of dangerously low oxygen levels.

Approach:

  • Leveraged scheduled Databricks notebooks to process terabytes of incoming data and deliver near real-time alerts.
  • Performed aggregated analytics (daily/weekly) and stored results in PostgreSQL for rapid access to frequently used metrics.

Outcome:

  • 95% of abnormal readings detected in real-time.
  • Significantly decreased query times for common metrics.
  • Scalable solution capable of handling terabytes of high-frequency data.

Think my experience aligns with your needs? or got any questions that I can help with? Let’s connect.

]]>Sagar DhungelDatabase Modernization From Mysql To Snowflake2025-12-20T00:00:00+00:002025-12-20T00:00:00+00:00https://sagrd.github.io/database-modernization-from-mysql-to-snowflakeA US-based lead generation agency was operating on an on-premise MS SQL Server originally built for transactional workloads, which had increasingly been used for reporting and analytics.

healthband er diagram

Challenge:

  • Limited scalability and high fixed infrastructure costs
  • Complex legacy schemas with undocumented SQL logic
  • Slow turnaround for new analytics and reporting needs

Approach:

  • Assessed existing SQL schemas, workloads, and critical reports
  • Designed a Snowflake-based data warehouse optimised for analytics
  • Migrated historical and incremental data from SQL Server to Snowflake
  • Refactored SQL logic into Snowflake-native transformations
  • Enabled role-based access and BI tool integration
  • Ran systems in parallel to validate data accuracy before cutover

Outcome:

  • Elastic, on-demand compute for reporting and ad-hoc analysis
  • Reduced infrastructure and licensing costs through a pay-per-use model
  • Faster and more reliable analytics delivery
  • Scalable foundation for advanced analytics and data science

Think my experience aligns with your needs? or got any questions that I can help with? Let’s connect.

]]>Sagar DhungelUnified Data Platform For Marketing Analytics2025-11-12T00:00:00+00:002025-11-12T00:00:00+00:00https://sagrd.github.io/unified-data-platform-for-marketing-analyticsAn entertainment agency faced fragmented and chaotic data across Social, Search, DSPs, and other third-party platforms.

healthband er diagram

Challenge:

  • Each platform (Meta, Google Ads, DV360, etc.) had its own schema, access method, and update schedule.
  • ETL processes were hard to scale, and difficult to monitor.
  • Business units waited days for performance insights due to disconnected tools and manual workflows.
  • Data scientists lacked a centralized workspace to share ad-hoc analyses and notebooks.
  • Ungoverned pipelines made audits and client reporting error-prone.

Approach:

  • Built a modern end-to-end data pipeline using Airflow and AWS-native tools, containerized with Docker for scalability.
  • Created a custom JupyterHub-based Data Science environment, enabling analysts to run, share, and collaborate on models and queries using unified warehouse data.

Outcome:

  • Faster, data-driven campaign optimization.
  • Analysts freed from manual data tasks, focusing on actionable insights.
  • Real-time visibility for marketing leaders.
  • Improved governance, audit readiness, and compliance.

Think my experience aligns with your needs? or got any questions that I can help with? Let’s connect.

]]>Sagar Dhungel