VANSHIKA GURBANI
********.*********@*****.*** LinkedIn Portfolio GitHub +1-551-***-**** New Brunswick, NJ EDUCATION
MS in Statistics - Data Science, Rutgers University - New Brunswick Sep 2023 - May 2025 Relevant Coursework: Probability & Inference, Database Management, Statistical Modeling Bachelor of Engineering in Computer Science, University of Mumbai Aug 2019 - May 2023 Relevant Coursework: DSA, OOPs, AI/ML, NLP, Deep Learning, Software Engineering TECHNICAL SKILLS
Languages: Python, R, SQL, JavaScript, TypeScript, Java, Bash Technologies: React.js, Next.js, Node.js, FastAPI, PostgreSQL, MySQL, LangGraph, LangChain, HuggingFace, Git, Docker, MongoDB, Tailwind, PHP, Git, AWS, GCP
Concepts: RAG Pipelines, LLM Agents, Prompt Engineering, Fine-tuning, Vector Embeddings, Multi-Agent Systems, NLP, ML Modeling, Full-Stack Deployment
EXPERIENCE
Software Engineer and Researcher - Particulate Systems Lab (PSL), Rutgers University Sep 2025 - Present
• Rebuilt pharmaceutical granulation simulations from MATLAB into Python ML pipelines, cutting processing time by 35% and validation time by 60% with automated benchmarking to confirm accuracy.
• TrainedGradientBoostingandRandomForestmodelson125simulationcasestopredictpharmaceuticalmanufacturing outcomes, replacing hours of manual simulation runs with instant predictions.
• ImplementedandmaintainedCI/CDworkflowsforactiveresearchtools,managingcodeupdates,environmentconfiguration, and deployment across GitHub Codespaces, JupyterHub, and SLURM-based HPC infrastructure servingresearchersand NSF program managers.
• AutomatedNSFgrantproposalsummarizationusingamulti-modelHuggingFacepipeline,eliminatingmanualsummarization for researchers reviewing large volumes of proposals. Software Engineer Intern - ContinuumCare.ai Oct 2025 - Nov 2025 (contract)
• Delivered the company’s health assistant MVP; patients upload medical records and ask questions, and the app returns AI answers grounded in 6+ clinical guidelines with cited sources.
• Enhanced response accuracy from20%to 90% through prompt optimization and guideline-based queryrephrasingusing a3-nodeLangGraphpipelinewithClaudeAPI,while reducing response timebyover60%(2minto45s)by implementing response streaming and result caching.
• Owned the full product build in Streamlit, including multi-user profile management, health record uploads, real-time AI summaries, and exportable PDF reports, all shipped within the contract period. Part-Time Lecturer & Teaching Assistant - Rutgers University Sep 2023 - May 2025
• Led weekly programming labs for 200+ undergraduates across 4 semesters; covered HTML, CSS, JavaScript, basic Excel
& SQL; earned 95% positive student feedback consistently.
• Redesigned course materials for a visually impaired student using accessibility-focused strategies, achieving strong performance relative to cohort benchmarks.
PROJECTS
SourceMD (link) (live) LangGraph, FastAPI, PostgreSQL, Docker, Render, Vercel, React, TypeScript, Tailwind
• Built a live AI medical fact-checker that scores AI-generated answers against real clinical guidelines on a 0-100 trust scale using a 3-node LangGraph pipeline with Groq Llama, Anthropic Claude fallback, Tavily web search, and ChromaDB over 844 guideline chunks across 4 medical domains.
• Deployed as a 4-service Docker stack across Render and Vercel with FastAPI, PostgreSQL, and a React TypeScript + Tailwind CSS frontend, with health monitoring, automatic restarts, and zero-touch redeployment on every code push. AI DevOps Agent (link) Next.js, Vercel AI SDK, Gemini 3.1 Pro, Supabase, GitHub API, Slack
• Built an AI-native DevOps agent at a hackathon hosted by Cerebral Valley, Vercel, and Google DeepMind, going from whiteboard to fully working demo in a single day.
• Monitors liveVerceldeployments,detects failures, scans for security issues, andautomaticallyopensGitHubpullrequests with fixes, integrated with Slack via slash commands for rollbacks, incident queries, and system status checks.