Hi! I’m Varsha

Machine Learning Engineer — National Aeronautics and Space Administration (NASA)

February 2026 – Present

• Optimize SEP prediction models by integrating heliophysics datasets, remote-sensing parameters from SDO, and realtime in-situ measurements from ACE/IMAP, improving F1-score by 15–20% and reducing false-alarm rates by 25%.
• Develop high-dimensional ML models to forecast post-solar-flare SEP events using datasets spanning Solar Cycles 23 and 24, achieving a True Skill Statistic (TSS) > 0.75 and improving average prediction lead time by ~45 minutes.

ORBIT CyberGraph Curation Intern — NJ Secure

February 2026 – Present

• Engineer a semantic knowledge graph in Neo4j to map 1,000+ global MISP threat feeds to MITRE ATT&CK techniques, creating a Digital Risk Twin that reduced LLM hallucinations by 40% through verified graph data grounding.
• Develop a Core Agent using the Model Context Protocol (MCP) and Chain-of-Thought logic to automate defense strategies for a state-wide platform, targeting a 100% successful pilot deployment across New Jersey community banks.

Full Stack Developer — AntiSnooze LLC

February 2026 – Present

• Architect a full-stack iOS ecosystem utilizing SwiftUI and CloudKit to manage real-time health telemetry; integration of wearable data contributed to a 25% increase in user retention through long-term behavioral habits and engagement.
• Engineer an ETL pipeline leveraging HealthKit APIs to analyze sleep-cycle data, addressing the 33% churn rate by delivering deterministic Smart Wake events that eliminate morning grogginess through sensor-driven optimization.

AI Research Intern — Siemens (AI Digital Twins Simulation)

May 2025 – Present

• Develop scalable software infrastructure for physics-informed neural PDE solvers using modular PyTorch classes.
• Containerize workflows with Apptainer for reproducibility across HPC environments; improve runtime by 38% via SLURM-based distributed training and I/O optimizations.
• Integrate graph-based simulation modules into Siemens’ industrial digital twin pipeline in agile sprints.

Machine Learning Researcher — XuLab (Brown/Giresun/NJIT)

Feb 2025 – Present

• Architect RGE-GCN: PyTorch-based GNN framework using Integrated Gradients and recursive elimination to identify cancer biomarkers.
• Develop a Streamlit web app integrating the framework into biomedical workflows.

Computational Biodesign Fellow — Grace Hopper-Kumar Lab

May 2025 – Aug 2025

• Architected containerized HPC workflows using Apptainer (Singularity) to deploy reproducible Python environments for generative diffusion models (RFdiffusion) on SLURM-managed clusters.
• Orchestrated high-throughput inference pipelines for protein structure prediction, utilizing SLURM job arrays to parallelize AntiBERTY and Graph Transformer models across distributed NVIDIA A100 GPU nodes.
• Automated massive-scale docking simulations by integrating DiffDock algorithms into SLURM batch scripts, optimizing GPU resource allocation for reverse diffusion processes over translations and torsions.

Intern & Instructor — Code Ninjas LLC

Jul 2021 – Aug 2024

• Mentored 100+ K–12 students through a comprehensive coding curriculum, teaching core computer science concepts (loops, conditionals, variables) using JavaScript, C#, and Lua.
• Debugged and troubleshot complex student projects in real-time, fostering critical thinking and resilience by guiding students through error analysis and logic correction.
• Facilitated technical workshops on game development fundamentals using the Unity engine, translating abstract programming theory into tangible, interactive results.