Goresh Sharma

Databricks Upgrade and Code Optimization for Precision Grid | Completed – 03/2025

• Optimized the "Precision Grid" and "Precision Area" codebase, upgrading from Databricks Runtime 10.4 to 14.3, and resolved dependency conflicts with libraries like Apache Sedona.
• Enhanced processing speed and resource efficiency by leveraging PySpark and Sedona’s latest features.

Gen AI Powered Sales Persona for Product Growth | Proof of Concept

• Conceptualized a multi-agent system using Generative AI (LLMs) to develop a dynamic sales persona that identifies customer needs, promotes product adoption, and personalizes interactions to increase revenue and market reach.
• Explored prompt engineering, agent orchestration, and knowledge retrieval techniques to create a highly adaptable sales assistant.

Open Route Service (ORS) Integration POC for Geospatial Projects | In Progress

• Leading a Proof of Concept to evaluate the integration of the Open Route Service (ORS) API into geospatial projects, focusing on routing, isochrone calculations, and network analysis.
• Assessing ORS accuracy, performance, and integration feasibility to enhance location-based services.

Precision Grids

• Spearheaded the development of "Precision Grids," a multi-resolution spatial framework integrating diverse input variables.
• Created population, prosperity, and sales indices to deliver actionable insights for strategic decision-making.
• Collaborated with technical teams to ensure seamless deployment, driving significant revenue growth and client portfolio expansion.
• Applied spatial imputation techniques to improve store estimate accuracy within each area.
• Utilized MLflow for dataset benchmarking of a worldwide grid, incorporating LandScan and VIIRS data to enhance model performance and reliability.

Precision Areas

• Optimized the "Precision Areas" product by transitioning the codebase from Python to PySpark and integrating machine learning for dynamic store estimation, enhancing functionality.
• Supported cross-functional teams in forecast-related projects, ensuring cohesive and impactful outcomes.

• Satellite Data Analysis:Developed an automated Python script to download, clean, and process Landsat satellite images for India monthly, computing spectral indices and storing results in an AWS database.
  
• USA Brands Data:Scraped and analyzed store location data for 300+ U.S. brands, performing geocoding and deduplication before integration into an Auto-ML platform as model inputs.
  
• Indian Brands Data:Collected and processed store location data for 200+ Indian brands, cleaning and analyzing it for use in Auto-ML models.
  
• PhonePe Pulse Data:Scraped and visualized PhonePe payment data using Python to support a client’s payment model.
  
• Population Data Assessment:Evaluated WorldPop raster images and Facebook’s VHRS images for global population data accuracy.