I build digital products that feel alive.

Sands Collections demanded a digital experience as refined as the products it showcases. The challenge was to build a blazing-fast e-commerce platform that could handle high-resolution imagery without sacrificing load times. The architecture leverages Next.js for server-side rendering, ensuring every product page is SEO-optimized and loads in under 2 seconds. A custom CMS integration allows the client to manage inventory, pricing, and promotional banners without touching code. Key decisions included implementing a headless commerce approach with Shopify as the backend, enabling flexible frontend design while maintaining robust order management. The checkout flow was optimized to reduce cart abandonment by 35% through a streamlined single-page process with real-time inventory validation.

SopWriter was born from a real pain point — students spending weeks crafting SOPs when AI could do the heavy lifting. The platform uses fine-tuned language models to generate university-specific statements that read naturally and hit all the right notes. The architecture is a Next.js frontend paired with a Python FastAPI backend. The AI pipeline uses prompt engineering with contextual templates for different universities, programs, and applicant backgrounds. Each generated SOP goes through a multi-pass refinement process that checks for tone, structure, and keyword optimization. User authentication is handled via NextAuth with Google and email providers. The payment system integrates Stripe for subscription management with tiered pricing. A built-in editor lets users refine AI-generated content with real-time grammar and style suggestions.

MoussenMelts is a local bakery that wanted its website to feel as warm and inviting as walking into the shop. The design prioritizes rich photography, warm color palettes, and smooth animations that make you smell the fresh bread through the screen. The menu system is dynamic — the owner can update daily specials, seasonal items, and pricing through a custom admin panel built with React. The ordering flow supports both pickup and delivery with time-slot selection and real-time availability checking. A standout feature is the "Build Your Box" customizer where customers can assemble gift boxes by dragging items into a virtual container with live price calculation. The entire experience is optimized for mobile since 70% of the bakery's customers order from their phones.

CreatubeAI tackles the bottleneck of video content creation for social media marketers. The platform takes a text prompt, selects relevant stock footage, generates voiceover, and assembles complete short-form videos — all in under 3 minutes. The video pipeline is orchestrated with a Python backend that coordinates multiple AI services: text-to-speech, scene matching, and automated editing. The rendering engine uses FFmpeg with custom presets optimized for TikTok, Instagram Reels, and YouTube Shorts aspect ratios. The Next.js frontend features a real-time preview system with WebSocket connections that stream rendering progress. Users can customize templates, brand colors, fonts, and music tracks. The dashboard includes analytics showing video performance metrics pulled from platform APIs. Scaling was the biggest challenge — rendering is GPU-intensive, so I implemented a job queue with Bull and Redis that dynamically allocates cloud GPU instances based on demand.