Back to Portfolio
Portfolio
12 projects

All Projects

Every project I have shipped, from enterprise AI pipelines and luxury retail experiences to open-source developer tools.

TallyMsg project cover
Next.jsNode.jsOpenAI APISaaSGenerative AISupabase
AI-parsed expenses via WhatsApp & Telegram

TallyMsg

A SaaS personal finance product I built end to end. Users log expenses by sending a plain message on Telegram or WhatsApp, and the AI parses the amount, merchant, and category automatically, with no app to download. A web dashboard shows charts, budgets, and spending insights, with tiered plans adding voice note logging, receipt scanning, and AI spending digests. I owned product, frontend, backend, AI, billing, and deployment.

Dawlati project cover
TypescriptNest.jsNode.jsPostgreSQL
99.9% uptime

Dawlati

A LinkedIn-style platform where UAE nationals hire Emirati individuals for a range of services, built under VentureDive. I led the backend as team lead for 4 developers, owning the repo, PR reviews, and work assignment while coordinating closely with stakeholders, UI, QA, mobile, and DevOps. I built the core matching engine, using AI to connect clients with the right Emirati professionals, along with Stripe-based payments, and worked with the DevOps team on production deployments that kept the platform at 99.9% uptime.

AskEddy project cover
TypescriptNest.jsNode.jsPostgreSQL
AI-powered bookings across UAE

AskEddy

An AI-powered platform for booking activities and attractions across the UAE, built at VentureDive. As backend lead for a team of 3 developers, I owned the repo, PR reviews, and work assignment, while coordinating closely with stakeholders, UI, QA, mobile, and DevOps. I built core features including payments, reporting, and the booking user journey, along with an AI recommendation engine that surfaces activities based on a user's interests and past bookings, plus the search that powers discovery across the platform.

Walmart Realm project cover
Next.jsReactChakra UIKRPano
Zero critical incidents

Walmart Realm

A gamified virtual shopping experience built for Walmart as part of an immersive retail platform. I built the Walmart Realm UI from scratch using React, Next.js, and Chakra UI, integrating third-party games and KRPano-powered 3D environments into a single interactive storefront. I also directed the end-to-end deployment pipeline across AWS and Google Cloud for a high-traffic, globally distributed audience, maintaining zero critical incidents through rigorous pre-deployment testing and proactive monitoring.

The Bicester Collection project cover
ReactChakra UIKRPanoAWS
Figma to live, full ownership

The Bicester Collection

Took the project from inception to production, building a responsive virtual shopping experience with React and Chakra UI from Figma designs. Integrated KRPano for the immersive 3D environment, added event tracking via Google Tag Manager, deployed on S3, managed CloudFront for delivery, and introduced gamification features to drive customer engagement.

Tommy Hilfiger project cover
ReactChakra UIKRPanoAWS
Gamification drove user engagement

Tommy Hilfiger

Led the project as the sole Front-End Engineer, building a responsive virtual experience with React and Chakra UI from Figma designs. Enhanced the KRPano-based experience by integrating the UI, adding event tracking via Google Tag Manager, and incorporating gamification elements to boost engagement. Managed full deployment via AWS S3 and CloudFront.

Afiniti Mega AI project cover
NestJSPostgreSQLRedisJest
90%+ unit test coverage

Afiniti Mega AI

Backend engineering on Afiniti's enterprise-grade AI behavioral analytics platform, serving high-throughput enterprise clients globally. I built and maintained scalable API servers using NestJS, PostgreSQL, and Redis, and pushed unit test coverage past 90% with Jest and Supertest, establishing a reliability baseline that significantly reduced production regressions. I also managed continuous deployment pipelines, coordinated zero-downtime production rollouts across a distributed engineering team, and integrated third-party APIs to extend platform capabilities.

Developer Toolkit project cover
ReactNext.jsTypeScriptPrivacy-First
100% browser-side, no signup

Developer Toolkit

A free, privacy-first collection of everyday developer utilities, including encode/decode, diff files, hash strings, and regex validation, with more tools added over time. Built with React, Next.js, and TypeScript, every tool runs entirely in the browser, so no file or string you paste ever leaves your machine. There is no sign-up, no data collection, and no rate limits, just fast, dependable utilities without handing data to a third party.

next-safe-env project cover
NPM PackageTypeScriptNext.jsEnvs
Catches misconfigs at build time

next-safe-env

An npm package that brings typed, validated environment variables to Next.js and Node.js projects. It validates your environment schema at build time instead of runtime, so misconfigured or missing variables get caught before deployment rather than causing failures in production. Full TypeScript inference means every environment variable is autocompleted and type-checked throughout your codebase.

Developer Toolkit MCP project cover
MCPTypeScriptClaudeCursor
Works with Claude, Cursor & Windsurf

Developer Toolkit MCP

An MCP server that gives AI coding assistants, including Claude, Cursor, Windsurf, and Zed, reliable, deterministic tools for everyday developer tasks: encode/decode, diff files, hash strings, decode tokens, and validate regexes. Rather than relying on an LLM to compute these operations itself and risk an inconsistent result, the assistant calls a dedicated tool and gets the same exact answer every time.

SageDesk project cover
TypeScriptRAGWebAssemblyBrowser-Native
Local-first, LLM-optional

SageDesk

A RAG-powered support chat widget that can be embedded into any website. It runs fully local in the browser via WebAssembly, so it works with no backend and no setup required. When richer, more conversational responses are needed, it can also connect to an LLM instead. Built with TypeScript, it is designed to be browser-native rather than a hosted widget calling out to a remote server.

AI Image Placement Pipeline for Virtual Showrooms project cover
PythonFastAPIAIPrompt EngineeringOpenCV
48-hr workflow cut to near-instant

AI Image Placement Pipeline for Virtual Showrooms

An end-to-end AI pipeline engineered with Python, FastAPI, and OpenCV that replaced a 48-hour manual 3D modeling workflow with a near-instantaneous alternative for e-commerce virtual showrooms. It takes a single product image and, using prompt engineering with predefined prompts, composites it onto a stand within 4K panoramic backgrounds. Built for global virtual commerce experiences at scale, it cut production timelines by 90%.