Data Scientist
Transforming raw data into actionable insights to improve healthcare outcomes, ensure rigorous compliance, and build resilient, enterprise-scale AI infrastructures.
Intuitive
Data Scientist (Co-Op) • Sunnyvale, CA • Jun 2025 - Mar 2026
Developed an Agentic AI Framework with Node.js, Langchain & Azure for 15+ Enterprise Apps, enabling rapid infusion of AI into enterprise workflows, resulting in a 65% reduction in support tickets.
Built robust RAG Pipelines for enterprise knowledge discovery, grounded AI with rigorous domain-specific datasets, and optimized ML infrastructure for high-throughput healthcare environments.
Data Governance & Healthcare
Data made available to AI compliant with HIPAA, with no PII leaving company servers, strictly following FDA and Medical-Devices Regulatory SOPs.
AI Evaluation Benchmarks
Constructed benchmarks to quantify AI performance on retrieval precision, grounding accuracy, hallucination rate & task success.
Inference Optimization
Reduced chat latency from 90 to 20 sec by tuning multi-threading for optimal GPU memory utilization & batching, yielding a 4x boost in RAG throughput. Minimized Azure & API costs using on-prem GPUs via Kubernetes.
University of Washington - IT
Software Engineer • Seattle, WA • Mar 2025 - Present
Architected massive-scale data pipelines on AWS (S3, Athena, Redshift) using Terraform to ingest and process org-wide digital assets, delivering actionable analytics to PowerBI dashboards.
Leveraged advanced MLOps to train gradient-boosted tree models (XGBoost, LightGBM) specifically targeted for intricate document structure analysis and remediation effort prediction. Furthermore, deployed advanced time-series forecasting models (ARIMA) to proactively project workload capacity and resource allocation across IT divisions.
AWS Data Architectures
Designed robust ELT pipelines processing websites, PDFs, and documents at organizational scale.
Automated Compliance
Automated WCAG 2.1 compliance evaluation by integrating Adobe APIs into pipelines to generate vital accessibility metrics.
Advanced Predictive Modeling
Trained XGBoost and LightGBM for complex document scoring and utilized ARIMA time-series models for precision workload and capacity forecasting.
Sequel2SQL
Microsoft Sponsored • University Capstone Project
Sequel2SQL is an agentic LLM + RAG framework specifically engineered for automated SQL error diagnosis, rigorous optimization, and self-correction. By deep diving into advanced SQL primitives—such as CTEs, Window Functions, and query optimization metrics—we developed highly nuanced reasoning methods for LLM and RAG performance.
By heavily scrutinizing low-level execution engine implementations, we were able to provide Language Models with highly targeted and semantically rich feedback. This enabled a 6% absolute improvement over the baseline on the rigorous BIRD-CRITIC Benchmark for Natural Language to SQL task performance.
AST Based Targeting
Designed AST-based, segment-level parsing and transformation layers to selectively process exceptionally large queries. This ensured peak memory efficiency while strictly applying highly targeted optimization strategies per query block.
Advanced SQL Execution
Explored and implemented the low-level mechanics of execution engines to accurately evaluate query planner paths, translating database-native heuristics into actionable reinforcement signals for LLMs.
Hypothesis Testing & A/B Tests
Conducted rigorous ablation studies on prompting, retrieval, and orchestration strategies, backing up architectural decisions with statistical hypothesis testing to identify the best configurations.
Education
University of Washington
Master of Science (MS), Data Science
Manipal Institute of Technology
B.Tech Data Science & Engineering
Minors in Finance & Portfolio Management