i like building software where technical depth and product design are treated as the same problem.
my work has moved across backend systems, developer tooling, computer vision, healthcare software, AI-assisted workflows, and mobile products and is usually somewhere between what a system does internally and what using it actually feels like.
| systems | product | signal |
|---|---|---|
| backend architecture | interface systems | observability |
| data modeling | interaction design | metrics |
| reliability | visual language | validation |
| performance | workflow design | testing |
numbers, but make them pretty ♡ · #050509 / #FF4FD8 / #63F58B / #7DD3FC
on-device ML recovery intelligence for iOS. HydraScan analyzes biomechanical movement using pose estimation and deterministic scoring, processing 33-point pose data at 120Hz and computing 50+ biomechanical metrics.
engineering notes / architecture
Tech: Swift, SwiftUI, MVVM, SwiftData, HealthKit, QuickPose, MediaPipe, SIMD3
- Scores squat, hip hinge, posture, balance, range of motion, and asymmetry.
- Uses NaN-safe joint-angle calculations and fallback logic for incomplete pose data.
- Built as a 12K+ line SwiftUI/MVVM client across 59 files.
- Includes a reusable 700+ line SwiftUI design system powering 20+ views.
- Designed as a privacy-first recovery app with deterministic scoring and on-device movement analysis.
repository-aware execution control for AI-assisted coding. BeliefGuard makes model assumptions explicit, grounds them against codebase evidence, and gates patch generation before workspace mutation.
engineering notes / architecture
Tech: TypeScript, VS Code Extension API, Zod, Vitest, LLM APIs
- Extracts AI agent assumptions into a typed Repo Belief Graph.
- Uses a Confidence-to-Action Gate to route changes to proceed, inspect, ask, or block.
- Integrates LLM plan extraction, workspace scanning, evidence grounding, Zod validation, structured patch generation, and per-file diff review.
- Validates behavior with Vitest, property-based tests, and benchmark scenarios.
- Built around responsible AI-assisted development, output validation, and developer trust.
industry-sponsored healthcare coordination platform for patients, caregivers, clinicians, and admins. Backend query improvements and pagination reduced large-dataset query time and network transfer by ~95%.
engineering notes / architecture
Tech: React, Vite, Node.js, Express, Prisma, PostgreSQL, Recharts
- Built role-based healthcare workflows across patients, caregivers, clinicians, and admins.
- Implemented RBAC, OTP verification, authenticated sessions, caregiver/MPOA linking, privacy controls, and audit visibility.
- Developed scheduling, messaging, medication tracking, vitals, notifications, feedback, and visit lifecycle workflows.
- Built admin analytics dashboards with KPI tracking, DAU monitoring, searchable logs, filters, and pagination.
- Optimized large-dataset query time and network transfer by 95% through pagination and backend query improvements.
swipeable civic discovery and action platform. A deployed civic engagement product that helps users discover, understand, and act on local policy issues.
Tech: Next.js, TypeScript, Prisma, PostgreSQL, Redis, Clerk, Claude
- Built feed, saved, profile, discussion, and issue-tracking workflows.
- Implemented Redis-backed state consistency and low-latency APIs.
- Built personalized ranking using interests, district matching, geography, and engagement velocity.
- Integrated Claude-powered summarization for dense civic issue content.
published AI/ML research project. A machine-learning project using Spotify API data to classify songs by emotional tone, reaching 76 percent classification accuracy.
Tech: Python, Keras, scikit-learn, Pandas, NumPy, Seaborn, Matplotlib, Spotify API
- Built a neural network for music emotion classification.
- Used feature preprocessing, scaling, label encoding, K-Fold cross validation, and confusion-matrix reporting.
- Published findings on acousticness, valence, tempo, and emotional tone classification.
| languages | systems + data | product | intelligence + data | quality + workflow |
|---|---|---|---|---|
| TypeScript | Node.js | React | LLM APIs | Git |
| JavaScript | Express | Next.js | Claude | GitHub |
| Python | REST APIs | Vite | Gemini | VS Code |
| Java | PostgreSQL | Tailwind CSS | Keras | Vitest |
| C++ | MySQL | SwiftUI | scikit-learn | Zod |
| Swift | Prisma | Pandas | JUnit | |
| SQL | Redis | NumPy |
full technology index
Languages — TypeScript, JavaScript, Python, Java, C++, Swift, SQL
Frontend / product — React, Next.js, Vite, Tailwind CSS, SwiftUI
Backend, APIs, databases — Node.js, Express, REST APIs, PostgreSQL, MySQL, Prisma, Redis
AI, ML, data — LLM APIs, Claude, Gemini, Keras, scikit-learn, Pandas, NumPy
Tools + workflow — Git, GitHub, VS Code, Vitest, Zod, JUnit
what i like building
- Full-stack applications with real users and clear workflows.
- AI-assisted tools that improve developer productivity and trust.
- Healthcare and clinical workflow software.
- Internal tools, dashboards, and automation systems.
- Data-driven applications with SQL-backed models.
- Frontend experiences that make complex systems easier to use.
- Reliable APIs, validation layers, and testable software.
what i'm going deeper on
- AI automation and LLM-powered workflows.
- Full-stack product engineering with React, TypeScript, and PostgreSQL.
- Backend API design and data modeling.
- Testing, validation, and software quality.
- Cloud fundamentals and deployment workflows.
- Healthcare, education, and developer tooling products.
- Workflow automation, observability, and reliable software systems.
| role | signal | |
|---|---|---|
| iDTech Camps | On-Campus Instructor | Java · 3D printing · character modeling · video · ethical AI |
| On My Own Technology | Research Intern | Python · Keras · Spotify API · music emotion classification |
| IndianRaga | Social Media Marketing Intern | Google Analytics · AdSense · YouTube · Instagram · Facebook |
experience notes
Taught Java programming, 3D printing, character modeling, video production, and ethical AI through hands-on student projects.
Researched music emotion classification and built a Keras/Spotify API machine learning model with Python and scikit-learn.
Analyzed Google Analytics, AdSense, YouTube, Instagram, and Facebook performance data to support growth and content optimization.
Computer Science graduate with a concentration in Software Engineering from Arizona State University.
My background spans full-stack software engineering, AI-assisted developer tools, healthcare technology, data-driven applications, and product-focused engineering. I enjoy building practical software that connects clean interfaces, reliable backend systems, structured data, and thoughtful user workflows.
original profile details, preserved
- Computer Science graduate with a concentration in Software Engineering from Arizona State University.
- Interested in full-stack engineering, frontend/product engineering, AI automation, healthcare software, internal tools, and data-driven systems.
- Strongest technologies: TypeScript, JavaScript, React, Node.js, Express, PostgreSQL, Python, Swift, SQL, Git, and REST APIs.
- I like building products end to end, from user flows and UI to APIs, databases, validation, testing, and documentation.
- Currently exploring AI-assisted development, LLM workflows, workflow automation, observability, and reliable software systems.
Primary Focus: Full-stack software engineering, AI tools, healthcare/product workflows
Frontend: React, Next.js, TypeScript, JavaScript, Tailwind CSS, SwiftUI
Backend: Node.js, Express, REST APIs, Prisma, PostgreSQL, MySQL, Redis
AI/ML: LLM APIs, Claude, Gemini, Keras, scikit-learn, Pandas, NumPy
Quality: Vitest, Zod, JUnit, validation, debugging, documentation
Interests: AI automation, healthtech, education technology, developer tools, internal tools
Legacy portfolio link from the previous profile: Website