Selected work
Web, software and AI projects shaped around real constraints.
Each case study explains the user problem, the product response and the technical choices behind the build. Browse the full project pages for scope, stack and next steps.
Case studies
V² Productions: A premium, cinematic creative studio website built for V² Productions with React 19, TypeScript, Vite, and Tailwind CSS. Showcases video editing, web design, digital marketing, live 1-on-1 courses, and dual mobile/desktop interactive UX. Architected a responsive dual-experience platform using React 19, TypeScript, Vite, and Framer Motion. Implemented mobile/desktop UI switching, animated hero showcases, lead gen flows, and modular course catalog structures.
KAMAYUU: Players can practice solo or create private multiplayer tables. The interface includes onboarding, player identity, tutorials, settings, score tracking, and table interactions, with card art, sound, and haptic feedback. Online play requires the configured backend. The project includes a dedicated rules engine, guided tutorial, card interface, and solo AI opponents. Online play uses Supabase authentication, realtime room communication, and a deployed game Edge Function.
STRIATUM 4.0: The site serves symposium delegates and organizers. Delegates can browse events, apply for access, select events, and submit payment proof. Organizers have registration-management screens. Event data and pricing rules are maintained in the repository. The site organizes symposium events into searchable categories and provides event details, registration forms, a cart, and payment-proof flows. Code supports Supabase-backed registration when configured, while the cart remains on the device. Payment is handled through submission of proof for review; it is not an automated card-payment flow.
Reelcut: Reelcut is being built for creators, colleges and small teams who record more video than they have time to edit. A creator uploads a long recording and the editor suggests cuts: removing silences and filler, generating word-timed captions, and reframing horizontal footage into vertical formats. Every suggestion stays editable on a timeline before export. Reelcut pairs speech transcription with a timeline editor. The transcript becomes the editing surface: deleting words removes the matching footage, silences can be cut in one step, captions are timed to each word, and vertical crops follow the speaker. AI proposes edits; the creator approves them.
Karaikal One: Karaikal One is designed for residents of Karaikal and nearby areas. It combines bus and train schedule browsing, local provider discovery, community help requests, lost-and-found posts and volunteer information. The interface supports English, Tamil and common Tanglish queries, and is built to work as a web app, installable PWA and future Capacitor-based Android application. Karaikal One provides a shared mobile-first experience for transport searches, local services, moderated community posts and an Ask Karaikal intent interface. The application labels demo, scheduled, unverified and community-reported information clearly so users can understand the limits of each result.
Neon Rail: Players run, change lanes, jump, slide, and avoid trains across three lanes. The game generates obstacle patterns under rules designed to keep them navigable. Players can use keyboard controls or enable webcam controls, including a head-only mode for seated play. Neon Rail combines a Three.js game with MediaPipe pose tracking in a Web Worker. It supports full-body and head-only camera controls, plus keyboard input; pose processing happens on-device, and the game makes no network requests after loading.
TEMPO: Players arrange four scrambled tiles into answer slots and submit an accepted word. The interface shows how many possible solutions remain, without displaying an unformed answer. It supports touch and keyboard input. Best score is stored on the device and is not synchronized. The app uses a deterministic game engine to track time, accepted words, score, and round state. Correct answers increase the tempo and add time; incorrect answers keep the same puzzle and award no points or time.
GSV OS: GSV OS is a private operating tool for BuiltbyGSV. It tracks work from lead through delivery and ongoing maintenance, with proposal versioning, requirements history, change-request handling, invoices, and a structured knowledge store. Optional AI can suggest actions, while the core workflows are designed to work without an AI provider. The platform organizes studio workflows into dedicated areas for pipeline, clients, proposals, projects, finance, maintenance, and structured memory. The repository documents a React and Supabase architecture, with business rules and transactional transitions enforced by the database.
Thaai Clinic: Thaai Clinic is a React + TypeScript healthcare app that simulates a full patient journey in a polished mobile interface. It includes onboarding, appointment booking, appointment tracking, medical record management, health education content, clinic details, and instant contact options through call and WhatsApp. Built a mobile-first prototype with simulated booking, local record management, sample health content and call/WhatsApp links. Backend integration and authentication remain future work.
- Web case study: https://www.builtbygsv.in/projects/v2-productions
- Software case study: https://www.builtbygsv.in/projects/kamayuu-card-game
- Web case study: https://www.builtbygsv.in/projects/striatum-4-symposium-platform
- AI case study: https://www.builtbygsv.in/projects/reelcut-ai-video-editor
- Web case study: https://www.builtbygsv.in/projects/karaikal-one
- Software case study: https://www.builtbygsv.in/projects/neon-rail
- Software case study: https://www.builtbygsv.in/projects/tempo-word-game
- Software case study: https://www.builtbygsv.in/projects/gsv-os-studio-operations
- Web case study: https://www.builtbygsv.in/projects/thaai-clinic-website
How to read the work
Projects include finished client work, active builds and BuiltbyGSV lab products. The status on each card distinguishes shipped work from concepts still being developed, while the case study records the current scope without presenting future features as complete.