Priyanka More

• Developed a physics-informed machine learning (PIML) digital twin in PyTorch for DC-DC boost converter parameter estimation, integrating neural networks with ODE-based simulation to achieve 5.19% MAPE on component prediction. • Engineered a synthetic data pipeline using SciPy ODE solvers and Latin Hypercube Sampling to generate 1,000 labeled waveform samples. Conducted observability analysis that reduced prediction error from 23% to 5.19%. • Benchmarked three modeling approaches - PIML, NARX-ANN, and feedforward networks, with the NARX predictor achieving R² = 0.9998 on open-loop waveform prediction.

• Organized 7+ cultural events for a community of 300+ international graduate students, managing end-to-end planning and day-of coordination. • Managed the club's social media presence, creating content and driving engagement to promote events and build community among international students at Stony Brook.

• Taught an intensive 6-week hybrid course introducing students to digital literacy, multimedia production, and data-driven storytelling. • Guided students through hands-on labs with Google Sheets (data visualization, pivot tables, dashboards), Clipchamp (video editing, audio/visual integration), and Notion (digital portfolios, resumes). • Designed and delivered interactive lectures, workshops, and real-world projects, blending technical and creative skills. • Mentored students in developing a final multimedia portfolio, showcasing their ability to analyze, design, and present digital content effectively.

• Developed and launched CodeHealth, an AI-powered code review platform using LLMs, Python, and React on AWS and Azure, analyzing 50,000+ lines of code and flagging vulnerabilities, reducing code quality issues by 15%. • Integrated OpenAI APIs with custom prompts to assess code quality, cutting manual review time by 2 hours per week across the engineering team. • Spearheaded the creation of 35+ configurable coding rules, enabling teams to customize linting and enforce their own coding standards. • Accelerated development cycles by 25% by integrating GitHub Copilot and Amazon CodeWhisperer into the SDLC. • Built automation test suites using Selenium and Java in early tenure, ensuring software reliability before transitioning to full-stack AI platform development.

• Developed backend systems using Python/Flask and frontend PWAs with Angular/Svelte, serving 1M+ users. • Built a recommendation engine using correlation analysis and clustering, along with a Random Forest Classifier to evaluate customer credit history and assign credit scores, improving creditworthiness evaluation speed by 18%. • Deployed Docker and Kubernetes configurations, enabling zero-downtime deployments and auto-scaling across production services. • Automated reconciliation of 2,000+ transaction discrepancies and performed root cause analysis to resolve data inconsistencies.

• Built a React-based web platform for the Cultivation Analysis project, developing interactive dashboards and charts to visualize crop and harvest data for agricultural planning. • Developed frontend UI components and backend API endpoints to support data retrieval and display, enabling farmers and stakeholders to track cultivation trends across seasons. • Followed Agile sprint workflows, collaborating with the team to iterate on features and deliver a responsive, user-friendly application.