experience

Education

CGPA 3.9/4.0. Graduate Assistantship, Aug 2024 – Jan 2026.

CGPA 3.1/4.0.

Research Experience

• Designed MAUQ-CLIP, the first black-box LLM uncertainty framework to treat missing evidence as an uncertainty signal, raising accuracy from 94.0% to 98.5% and cutting calibration error by 88% on SEER uterine cancer records.

• Ran the held-out contamination experiment, deleting the facts that generated half the benchmark claims and re-scoring to separate reasoning gains from retrieval of the answer key.
• Served as one of two annotators on a 110-claim clinical label review (Cohen’s kappa 0.966), identifying 6.4% of labels as context-dependent.

• Co-developed GANterpolate, a hybrid U-Net GAN and interpolation framework cutting temperature RMSE from 6.84 to 5.37 on ERA5, then extended it (GANterpolate+) across agriculture, energy, healthcare and finance, beating both baselines in all five domains.

Teaching and Professional Experience

• Supporting CSIT 598 Machine Learning, CSIT 559 Generative AI, CSIT 696 Research Methods and CSIT 697 Master’s Project.

• Developing CoChair, an AI assistant in the Gmail sidebar that helps faculty triage student requests, retrieve the applicable institutional policies, and draft recommended replies; piloted with the Biology Department.
• Improved policy retrieval over a Qdrant vector store (email-as-query search, relative selection, precedence and recency rules), built reply redrafting on faculty overrides, and built the policy management screen on Firestore.

• Taught 450+ students across 15 course sections; authored the laboratory assignments, held 20 office hours weekly, and reported performance trends to faculty to inform grading standards.

• Built Client Central, an internal onboarding platform, cutting onboarding initialization time by 60%. React and TypeScript front end, Java and Spring Boot microservices with Kafka; owned CI/CD across AWS, Jenkins and Docker.

Honors and Awards

Certifications

All credentials are verifiable on Credly.

Technical Skills

• ML and LLM systems: Python, PyTorch, TensorFlow, scikit-learn; retrieval-augmented generation (LangChain, LlamaIndex, Qdrant); generative models (GAN, TimeGAN)
• Engineering: Java, TypeScript, SQL; AWS (Bedrock, SageMaker, EC2, S3), Google Cloud (Cloud Run, Vertex AI, Firestore), Docker, Kafka, PostgreSQL, React, Spring Boot