Experience

Graduate Research Assistant

TReNDS Center, Georgia State University · Atlanta, GA

• Lead my doctoral research on multimodal medical AI that holds up when data is incomplete: modality-aware perception, latent disentanglement under missing modalities, and longitudinal clinical reasoning that keeps track of uncertainty.
• Developed a probabilistic multimodal framework (HSIC-constrained latent factorization plus a Self-Critiquing Mixture-of-Experts) that set state of the art across 7 systemic and ocular conditions.
• Developed uncertainty-aware longitudinal reasoning that treats disease progression as distributional change on statistical manifolds (Fisher-Rao), so diagnostic ambiguity survives and claims stay tied to image evidence; state of the art with +7.4% CE-F1 on Longitudinal MIMIC-CXR (Dist-RRG, STRIDE, DELTA).

Machine Learning Engineer II

UST Global Inc., AI Center of Excellence · Remote · Aliso Viejo, CA

• Lead a 21-person AI team in the AI-CoE building UST Alpha AI, UST's one-stop shop for AI workloads, and its platform-modernization tools. Led development of CodeCrafter; designed and built Orbit.
• Run delivery on an AI-DLC approach: agents migrate legacy .NET 4→8 and COBOL monoliths to React/Java Spring Boot, generating NX monorepos from specs that land 95% functional, with delivery 70% faster at 60% of the cost. Now reimagining a NextGen Reinsurance Platform with SpecKit; the MVP already works.
• Built PathForce, an AI-assisted anatomic-pathology LIS (rules-first eligibility checks, effort-weighted case distribution, HL7/FHIR/X12, Part 11-grade audit trail), plus RiskIQ for ICD-10 coding and risk adjustment, and PolicyIQ, which turns CMS/payer policy into SME-approved claim edits in minutes, not weeks.
• Modernized a payer's claims adjudication by re-engineering its extracts and reporting, improving payment accuracy and audit-readiness. Architected and led Call Analytics Agents that raise First-Call Resolution, reduce repeat contacts, and flag quality and compliance issues.
• Architected BSA DataMap (agentic profiling into BigQuery, saving 2 FTEs/$250K) and an ETL Test Agent validated with LLM-as-a-Judge. Delivered RegView, natural-language search over regulatory documents, and a RAG-powered RFI/RFP engine that cut proposal time by 60%.
• Collaborated with Dr. Tatsunori Hashimoto (Stanford AI Lab) on AxisSQL, an inference-time-scaling Text-to-SQL framework ranked #12 on the BIRD leaderboard at submission; submitted to EACL (Industry Track).

Graduate Research Assistant

TReNDS Center, Georgia State University · Atlanta, GA

• Customized multimodal LLMs in PyTorch to generate reports from medical images and text: 13.4% higher BLEU-4 than VisionGPT, with inference down to 1.6 s per image.
• Built TensorFlow/Keras image classifiers for retinopathy, schizophrenia, breast cancer and colon cancer, cutting false negatives by 5–7%.
• Multi-modal Imaging Genomics Transformer: fused genomics with sMRI and fMRI, improving schizophrenia diagnosis accuracy by 2.12%.
• Guided Context Gating and Spatial Sequence Attention: attention mechanisms for retinal images and structural brain MRI.
• Medical Vision Language Transformer: an Abstractor-Adaptor framework for resource-constrained medical image captioning.

Software Engineer III

Carelon Global Solutions (Elevance Health) · Hyderabad, India

• Built REST APIs for the COmpensation INcentive System, a microservices application that validates, computes and speeds up incentive payments, cutting processing time by 10%. Wrote Python/SQL clean-up scripts that stopped over $500K in overpayments.
• Fixed critical production issues in the commissions calculation flow (agent advance computation), saving $1.5 million in historical overpayments; earned the Go Above – Impact Award.

Systems Engineer, Analytics & Insights

Tata Consultancy Services · Hyderabad, India

• Designed a log-based Bug Root Cause Prediction System with an Attention LSTM in Azure ML Studio, cutting debugging effort and saving 3 full-time equivalents a year.
• Set up an Azure DevOps model-deployment pipeline that took deployment from 2 to 1.2 hours and raised system availability by 25%; named team Star Performer.
• Optimized order-orchestration data pipelines, reducing fallouts by 30%, and built operational dashboards for data-driven decisions.

Ph.D. in Computer Science

Georgia State University · Atlanta, GA · GPA 4.23 / 4.30

Dissertation: From Perception to Clinical Reasoning (see Research).

Coursework: Advanced Machine Learning, Deep Learning, Computer Vision, Natural Language Processing, Digital Image Processing, Advanced Image Processing, Computational Intelligence, Privacy Aware Computing, Fundamentals of Data Science

M.S. in Computer Science (Thesis)

Georgia State University · Atlanta, GA · GPA 4.17 / 4.30

Thesis: Attentive Multi-modal Learning (see Research).

Coursework: Advanced Machine Learning, Deep Learning, Computer Vision, Natural Language Processing, Digital Image Processing, Computational Intelligence, Fundamentals of Data Science

B.Tech. (Honours) in Computer Science & Engineering

Vignan's Foundation for Science, Technology & Research · Guntur, India · GPA 9.79 / 10 (3.92 / 4.0)

Chairman's Gold Medal; Best Outgoing Student of CSE.

Coursework: Data Structures, Design & Analysis of Algorithms, Probability & Statistics, Artificial Intelligence, Pattern Recognition, Data Mining, Optimization Techniques