About Me

AI Engineer at Daiichi Sankyo

As an AI Engineer Intern, I built enterprise-grade Generative AI systems for large-scale medical and research intelligence. To tackle complex clinical reasoning, I architected a Hybrid Graph RAG Engine that synthesizes unstructured corpora into interconnected knowledge graphs, enabling multi-hop reasoning across disjoint documents. I paired this with aggressive optimization by fine-tuning 4-bit quantized GPT-OSS-20B using QLoRA. This reduced inference latency by 60% and lowered GPU compute costs, enabling faster, cost-effective analysis without compromising response fidelity.

I engineered a context-aware semantic chunking algorithm tailored to preserve meaning in dense medical text. I integrated it with a GPU-accelerated FAISS vector search pipeline and NetworkX graph traversal, using MMR reranking to improve retrieval diversity and precision. This pipeline significantly improved the system’s ability to surface specific clinical evidence buried across thousands of documents.

For production readiness, I deployed the solution as a high-throughput RESTful microservice architecture capable of ingesting and sanitizing diverse datasets with a 50% reduction in processing latency. To meet strict healthcare compliance requirements, I integrated multi-layer AI guardrails and automated CI/CD pipelines for cloud-agnostic deployment across AWS and Google Cloud. This ensured 99.9% system reliability and zero-trust privacy compliance, delivering secure, scalable AI infrastructure for the organization.

In parallel, I built a data integration application that ingested medical records daily from the VVRIM application via APIs, transformed them, and stored structured outputs in Amazon Redshift. Using CRON scheduling, the pipeline automated ingestion and transformation while preserving historical consistency by recreating past datasets for any given date, enabling reliable real-time and historical analytics.

Research and Teaching Assistant at UB

My research centers on Green AI through low-level systems optimization. I engineered high-performance C++ and Python kernels, implementing SIMD vectorization and multi-threaded synchronization. By refining hardware-level data access patterns, I achieved a 15% reduction in energy consumption across distributed deep learning workloads without compromising computational speed.

To validate these improvements, I executed rigorous profiling workflows using Linux tools such as perf and Valgrind, identifying runtime inefficiencies including L3 cache misses, lock contention, and memory leaks. This diagnostic work enabled refactors that improved memory locality and thread safety, maximizing hardware utilization for large-scale AI workloads.

In parallel, I serve as a Graduate Course Assistant for the Machine Learning course. I mentor students on algorithmic concepts, provide structured feedback on projects, and guide optimization-oriented improvements, helping teams translate theory into clean, high-quality implementations.

Software Development Engineer at Experian

I engineered a high-throughput Java Spring Boot microservices architecture built for the operational demands of the financial sector. Leveraging Apache Kafka, I implemented asynchronous event streaming patterns to process millions of credit records daily. This fault-tolerant design decoupled ingestion from downstream processing, improving availability and preventing resource overload during peak traffic.

For the CPD Waterfall decision engine, I implemented multi-threaded parallel processing to improve throughput under heavy concurrency. I diagnosed and resolved race conditions and batch execution inefficiencies, introducing optimized synchronization strategies that reduced report generation time by 35%, enabling consistent performance for high-volume concurrent users.

I targeted bottlenecks at the data layer by refining sharding strategies and analyzing query execution plans for massive datasets. By resolving critical lock contention and IOPS bottlenecks within MySQL, I improved overall system response latency by 25% and strengthened scalability under load.

Beyond core performance, I built RESTful API aggregators to orchestrate secure data retrieval across internal services, creating a standardized integration layer that reduced network overhead and improved response times. I also contributed to platform modernization by migrating a legacy stack from Spring 4 to Spring 5, reducing technical debt and strengthening security posture while enabling a more modular, cloud-ready architecture.

SDE Internship at Experian

As a Software Development Engineer Intern, I focused on backend performance improvements for the CS-Access application, a critical system for credit score management. I diagnosed and resolved severe N+1 query bottlenecks by implementing advanced Hibernate caching strategies. This optimization decreased database load and reduced user wait times by 25%, ensuring the application could handle peak concurrent access without latency degradation.

To handle high-volume operations, I orchestrated asynchronous batch workflows using Quartz Scheduler. I designed these jobs with memory-efficient execution strategies to process large datasets in the background, preventing system resource overloads during business hours and maintaining stable throughput for recurring credit score updates and external API integrations.

Prioritizing stability and automation, I engineered comprehensive JUnit test suites using Mockito to simulate complex financial scenarios, increasing code coverage by 40% and preventing regressions. I supported this with a robust CI/CD pipeline using Jenkins and automated regression suites, reducing software release cycle overhead by 80%.

Beyond core deliverables, I led a hackathon initiative to architect a Transactions Application using Spring Boot. The tool gave production teams real-time access to subcode and transaction data, accelerating ticket resolution and improving operational visibility across support workflows.

Research Assistant at KMIT

During my research internship at the Research and Development Lab of Keshav Memorial Institute of Technology, I actively contributed to advancing medical imaging analysis, focusing on breast carcinoma pathology. I developed and refined AI-based deep learning models to detect tubules in whole-slide images, enhancing the accuracy of medical diagnostics. Leveraging frameworks like Faster R-CNN and YOLOv5, I designed a robust pipeline to identify mitosis in digitized pathology images, significantly aiding in breast cancer detection. I demonstrated expertise in utilizing tools such as QuPath to construct systematic workflows for annotating Whole Slide Imaging (WSI) TIFF files, producing high-quality images and masks for accurate labeling. Collaborating within a four-member team, I effectively handled image preprocessing, implemented distributed training strategies, and showcased strong leadership and presentation skills. These efforts highlighted my ability to tackle complex challenges in the intersection of AI and medical research.