Jan 2023 - Jun 2023
BISAG-N
Software Engineer Intern
Built backend and database components for a forest-fire prediction research platform, focusing on PostgreSQL data modeling, Flask API optimization, research analytics, and map-based dashboard support.
Role Overview
A quick read on what the role was and the main areas I worked on before the deeper implementation details.
Built the backend data layer for a forest-fire prediction research platform, transforming 100K+ daily geospatial and environmental records into structured PostgreSQL datasets for model training, prediction review, and dashboard analytics.
Redesigned PostgreSQL tables around location, timestamp, weather attributes, fire-risk scores, and model outputs, adding query-focused indexes to support faster filtering, aggregation, and comparison across regions and time periods.
Optimized Flask API query paths by moving filtering and aggregation logic into PostgreSQL instead of repeatedly processing large datasets in the backend, improving dashboard data access performance by 10-15%.
Connected optimized backend APIs with a Next.js, React, and Google Maps dashboard, enabling researchers to inspect high-risk zones, compare prediction outputs, and reduce repeated infrastructure usage by 6-8%.
Detailed Work
The role is grouped by major workstreams, with each one kept concise enough to explain the implementation and impact without turning into a wall of text.
Backend Data Layer for Forest-Fire Research
Built the structured backend data layer that turned large raw geospatial and environmental datasets into queryable PostgreSQL data for prediction review, analysis, and dashboard use.
Built the backend data layer for a forest-fire prediction research platform, transforming 100K+ daily geospatial and environmental records into structured PostgreSQL datasets for model training, prediction review, and dashboard analytics.
Organized processed data around core research dimensions such as location, timestamp, weather attributes, fire-risk score, and model output so researchers could compare results without repeatedly handling raw files in the backend.
This gave the research workflow a more queryable and reusable foundation instead of leaving prediction data trapped inside one-off processing steps.
PostgreSQL Query Design and API Optimization
Redesigned the database and API query paths so filtering and aggregation happened in PostgreSQL instead of repeatedly moving large datasets into Python for backend-side processing.
Redesigned PostgreSQL tables around location, timestamp, weather attributes, fire-risk scores, and model outputs, then added query-focused indexes on the fields most frequently used in filtering and comparison.
Optimized Flask API query paths by pushing filtering and aggregation into PostgreSQL instead of repeatedly fetching large datasets into Python and processing them in the backend.
This improved dashboard data access performance by around 10-15% by reducing repeated large scans, unnecessary data transfer, and inefficient application-side query handling.
Research Dashboard and Infrastructure Efficiency
Connected the optimized backend to a map-based research dashboard so prediction outputs could be reviewed visually while the data pipeline avoided repeated processing and unnecessary infrastructure use.
Connected optimized Flask APIs to a Next.js, React, and Google Maps dashboard so researchers could inspect high-risk zones, compare prediction outputs, and analyze fire-risk patterns visually by location and time.
Supported workflows handling 1-2GB of daily research data across 100-300 processing jobs, with raw, processed, and analytics-ready data kept separate to reduce repeated work.
This helped reduce infrastructure usage by around 6-8% while making the dashboard workflow more reliable for day-to-day research analysis.