Spearheaded the end-to-end migration of mission-critical ETL pipelines from PostgreSQL (Greenplum) to SparkSQL and PySpark on Databricks, improving job runtimes by 80% via logical flow enhancements.
Owned the development of backend architecture on Databricks for 70+ Business Objects (BO) reports integrated with Oracle ERP systems, streamlining complex reporting for key business stakeholders.
Maintained the GSC Bowler (Global Supply Chain) performance-tracking system, ensuring the integrity of high-fidelity data for critical supply chain KPIs.
Modernized the Enterprise Data Lake (EDL) by transitioning sources from PowerNow 1.0 to 2.0, ensuring architectural consistency and zero data loss for global operations.
Orchestrated robust workflows using Apache Airflow and Databricks native features to ensure scalable, production-grade data processing.
Engineered Gen-AI and RAG systems using Python to automate insight generation and improve data discoverability within the data ecosystem.
Collaborated across cross-functional teams to streamline migration across Ingestion, Transformation, and Visualization layers.
Developed a Python automation script using LangChain to generate summaries of 10-K financial reports, achieving an 85% speedup over existing manual solutions.
Built a Financial Health Dashboard for Consumer Banking using Power BI, providing a comprehensive overview of income and spending with automated goal suggestions.
Performed data preprocessing and cleaning to extract actionable insights for banking sector stakeholders.
Intern – Android Development & Security
DRDO
New Delhi, DelhiJul 2023 – Aug 2023
Developed a background Android application using Kotlin and Jetpack Compose to capture and forward system notifications via SMS.
Conducted security analysis using MobSF (static and dynamic), identifying critical flaws and reducing total testing time by 70%.
Implemented real-time logging and system-level API integrations (SmsManager) for secure notification monitoring.
Projects
What I've Built
Loocle – Self-Hosted Photo Backup & Web App
Engineered a cloud-independent photo management system replicating the Google Photos experience on local hardware.
Built a Python Flask web app with fast metadata-based search, caching, and multi-shot facial recognition features.
Automated wireless Android backups using ADB over Wi-Fi for seamless syncing between devices.
Implemented automatic metadata tagging using local LLMs via Ollama to associate detected faces with person names, enhancing organization and quick retrieval.
Face Recognition via One-Shot Classification
Recognizes faces from a single source image using embedding differences.
Over 90% speedup versus traditional face recognition.
Built with FaceNet, Haar Cascade, and one-shot classification.