Keyword search fails to capture context in large PDF datasets, making information retrieval inefficient.
Built a semantic search engine using OCR and vector embeddings to understand document meaning.
Manual chart interpretation is slow and lacks metadata.
Multimodal pipeline using YOLOv5, OCR, and LLMs.
Deepfake generation required complex tools and expert knowledge.
Motion transfer framework via First Order Motion Model.
High buffering times and latency during media playback.
Developed a RESTful server reducing buffering by 87% via range requests and caching.
Led AI-based monitoring and incident response workflows, significantly reducing mean time to detection. Designed and deployed automated document processors using OCR and LLMs, achieving 93% precision. Integrated MongoDB to support searchable docs and leveraged Power BI for cost optimization.
Developed a unified Power BI dashboard using a star schema model to track revenue and profit, uncovering insights that drove regional fulfillment strategies.
Improved API throughput by 4x and reduced response time to 20ms using Django caching and asynchronous logic. Developed RESTful APIs to validate and search documentation. Increased test coverage to 96% via unit testing and automated API testing within an Azure environment.
