Shrey Sawant - Vibe Coding | Contra
Work by Shrey Sawant
Sign Up
Post a job
Sign Up
Log In
Shrey Sawant
Harvard CS50 grad. I build clean, responsive ai websites.
Message
Follow
New to Contra
Shrey is ready for their next project!
Mumbai, India
Work
Posts
Services
About
Mumbai, India
2
I made a few changes in the project and fixed some of the bugs that were there during my prompting and image description
2
65
0
Ambika Interior Website Project This project is a premium, portfolio-led website for Ambika Interior, Malad East, Mumbai. The visual direction is inspired by the reference website Studio Anton, while the branding, copy, layout and imagery are original. Business information used Business: Ambika Interior / अंबिका इंटीरियर Category: Interior designer Google rating: 4.8 from 20 reviews Phone: 093246 20588 Address: Shop No. 8, Ratnaprabha, Shivaji Chowk, Plot No. 705, Western Express Highway, Dudhsagar Society, Tanji Nagar, Kurar Village, Malad East, Mumbai 400097 Services positioned: Residential interiors, modular kitchens, bedrooms, wardrobes, living spaces and workspaces Website sections Hero section “Spaces that feel like you.” Mumbai-focused positioning Explore Our Work CTA Visit Our Studio CTA Google rating display About section Personalised design philosophy “Good design begins with listening” Trust-building statistics Services section Complete home interiors Modular kitchens Bedrooms and wardrobes Living and workspaces Projects gallery Editorial-style project cards Hover interactions Responsive asymmetrical grid Original interior design imagery Design process We listen We shape We make Testimonials Interactive testimonial carousel Client names and project context Contact section Direct phone CTA WhatsApp CTA Studio address Business hours Instagram link placeholder Mobile experience Responsive mobile layout Mobile navigation menu Floating WhatsApp button Mobile-optimised typography and spacing Design direction Mood: Warm, calm, premium and architectural Colours: Deep forest green, warm ivory, sand, walnut brown and muted gold Typography: Playfair Display for editorial headlines and DM Sans for body copy Style: High-end editorial portfolio rather than a generic local-business template Interactions: Smooth scrolling, hover transitions, testimonial controls, responsive navigation and CTA micro-interactions Copyright approach Photos were not copied from Google Maps, Instagram or any third-party website. The website uses original AI-generated interior visuals. Google Maps was used only to verify business information, location, rating and contact details. Technical status Built with React, TypeScript and Tailwind Responsive desktop and mobile layouts Production build successfully verified Project checkpoint saved and ready for further edits Open website (manus-webdev://f17125f9) Details that may still need to be added or replaced Actual completed-project photographs Official Instagram URL Business email address Confirmed working hours Enquiry/contact form Verified client testimonials Existing logo and official brand colours, if available
0
16
0
PhysiqueForge is a premium, dark-themed AI fitness platform inspired by the high-energy aesthetic of School of Elites. It is designed to guide a user from onboarding through personalized training, nutrition, physique targeting, and ongoing coaching. Main user journey Landing page A high-contrast fitness campaign page with a cinematic hero visual, animated statistics, feature sections, comparison content, testimonials, FAQs, and calls to begin the assessment. 50-question assessment An exact 50-question, multi-step questionnaire covering body measurements, lifestyle, training history, diet, recovery, preferences, and goals. It includes progress tracking and saves the user's answers to their profile. Target calibration Users can choose a preset physique style or describe their desired look. The system maps that goal to a realistic target based on the user's assessment rather than promising an identical or guaranteed transformation. 84-day plan generation The platform generates a personalized 84-day training plan with daily sessions, compound-lift emphasis, progressive overload, recovery guidance, and nutrition targets. Two distinct guidance paths — Users choose either a Natural Path, focused on whole foods and non-supplement strategies, or a Supplement Path, which can include general guidance around products such as protein and creatine. These paths are mutually exclusive in the data model and UI. Member dashboard The dashboard shows the active plan, current day, phase overview, today's workout, nutrition targets, workout logging, and progress over time. Jimm AI coach Jimm is the platform's personalized fitness and nutrition assistant. It uses the user's saved assessment, target physique, active 84-day plan, and previous chat history as context.
0
22
0
OMNI AI Employee OMNI is a startup MVP for an AI employee workspace. Its positioning is: Turn complex, high-context work into structured plans, decisions, and next actions. It is designed for founders, operators, consultants, and small teams who want to delegate research, planning, and operational work to an AI system. What is included Public marketing site The homepage includes: Product positioning and value proposition AI employee demo visualization “How it works” workflow Pricing tiers: Starter — $49/month Pro — $199/month Enterprise — Custom Responsive mobile navigation Calls to launch the workspace or sign in AI workspace The /app workspace includes: Overview dashboard AI task input area Suggested prompts such as: Competitor research Meeting-to-action conversion Customer signal analysis Live AI execution Structured results containing: Summary Recommended sequence Next actions Estimated time saved Confidence level Project management Users can: Create projects Add project descriptions View project progress Organize recurring work into project spaces The database includes persistent projects and runs tables. Activity history Completed AI runs are recorded with: Original prompt Execution status Summary/result Timestamp Associated user/project data Workflow library The workspace includes starter workflows for: Competitor intelligence Meeting notes to execution plans Customer feedback analysis Authentication The project is scaffolded with Manus OAuth support, including: Sign-in flow User session handling Sign-out User-aware workspace display Database-backed user records Technical architecture Frontend: React, TypeScript, Vite, Tailwind CSS Backend: Express and tRPC Database: MySQL/TiDB with Drizzle ORM AI: Server-side built-in LLM integration Authentication: Manus OAuth UI: Responsive dashboard plus public marketing site Testing: Vitest Deployment model: Managed WebDev project with a running development preview The main server procedures include: projects.list projects.create activity.list ai.execute auth.me (http://auth.me) auth.logout Important current limitations This is a strong working MVP, but it is not yet a complete commercial SaaS business. The following parts are currently represented as product scaffolding or placeholders: Stripe billing is not connected yet. Notion, Slack, Google Drive, Linear, and HubSpot connectors are visual placeholders. Workflow customization is not implemented yet. Team permissions and approval workflows are not implemented. The AI can generate plans and recommendations, but it does not yet autonomously browse external services or modify third-party systems. The pricing tiers are marketing UI only until billing is added. The AI execution includes a fallback response, so the workspace remains usable even if the model call fails. Verification completed The project passed: TypeScript validation Production build Existing authentication test New OMNI procedure tests Desktop visual verification Mobile responsive verification Current checkpoint: Open the OMNI project (manus-webdev://652eaf15) Best next development phase The most valuable next phase would be: Add Stripe subscriptions and usage limits Add one real integration, preferably Notion or Slack Add team workspaces and invite-based collaboration Add an execution timeline showing AI planning, research, and completion states Create a real onboarding flow that guides users through their first task
0
24
Challenges
Lovable Challenge
Rive Halloween Challenge
The invideo Editor Challenge
Challenges
Lovable Challenge
Rive Halloween Challenge
The invideo Editor Challenge