Mar 2024 – Jul 2025 · 1 yr 4 mos

• Designed and deployed a Neo4j-based Knowledge Graph for the OMICS Biological Database, integrating heterogeneous biological data via BioCypher within a Dockerized microservice architecture to ensure scalability and modular data management.
• Architected a neuro-symbolic RAG pipeline that fuses LLM-based generation with knowledge graph reasoning for the OMICS domain, enhancing factual grounding and reducing hallucinations in biological data interpretation.
• Developed an intelligent NLP interface using Flask and the fine-tuned LLM, allowing researchers to query the knowledge graph in natural language and receive contextually enriched, generated insights — improving interpretability and discovery efficiency.

Jan 2022 – Mar 2024 · 2 yrs 2 mos

• Engineered advanced deep learning ensemble architectures to predict protein mutations at specific genomic loci, significantly enhancing mutation prediction accuracy and accelerating computational biology research pipelines.
• Deployed end-to-end predictive models and built an interactive analytics dashboard, empowering researchers with real-time visualization and interpretability of protein mutation insights through an intuitive, in-house interface.
• Co-developed an internal intelligent chatbot leveraging NLP and knowledge retrieval systems to automate biological data queries and streamline cross-functional communication workflows.
• Contributed to a Generative AI–driven entity extraction framework for medical claim datasets, improving semantic understanding, data normalization, and scalability within healthcare data processing pipelines.

Jan 2021 – Jan 2022 · 1 yr

• Developed a deep learning–based computer vision model to detect heat signatures from UAV drone imagery, enabling temperature estimation and identification of potential combustion substances for proactive hazard detection.
• Optimized a face detection system by refactoring legacy code and building an end-to-end ML pipeline spanning data ingestion, preprocessing, and model deployment — improving inference speed and overall system reliability.

Jul 2020 – Nov 2020 · 4 mos

• Developed a chatbot using the RASA framework to deliver real-time product status updates on Cashify’s buy-and-sell platform, reducing customer service calls by 30% and improving response efficiency.
• Built a dynamic pricing model for used smartphones using multiple regression and Random Forest algorithms, achieving an R² score of 0.91 and enhancing price accuracy by 18%.
• Applied transfer learning to an object detection model for smartphones and refined object boundaries using OpenCV’s Convex-Hull algorithm, improving detection precision by 22%.