Arslan Ahmad

I worked on Assignment Desk, a B2B marketplace connecting production companies like UFC, NFL, and Pepsi with freelance media crews. I built features covering the full workflow from talent search and booking through to final billing, and integrated Stripe for marketplace payments between clients and vendors. The directory had over 4,000 professionals with advanced filtering — I managed that data in MySQL and handled complex state in React and Redux across multi-step booking flows.

I developed a specialized HR portal at Nextek Solutions to manage and automate corporate gifting workflows for enterprise employees. I architected and built the complete back end using Node.js, Express, and TypeScript, ensuring strict type safety and maintainable code throughout. MongoDB was used for flexible, scalable storage of employee records and gift inventories. I integrated SendGrid for automated email notifications, triggering personalized gift confirmation and dispatch alerts to employees. The back end was deployed on AWS EC2 with a fully automated CI/CD pipeline via GitHub Actions, streamlining testing and deployment while eliminating manual overhead. I collaborated closely with front-end developers to integrate a Vue.js interface with the back end through secure RESTful APIs, ensuring seamless data flow across the full stack.

I led the development of a high-concurrency automotive auction platform serving the UAE market. I implemented real-time bidding using Socket.io, enabling instant bid updates across all connected clients with sub-second latency for international users. The back end is an Express.js server containerized with Docker and deployed on AWS ECS with ECR as the container registry. The front end consists of two separate React portals — a member-facing bidding portal and an admin management portal — both deployed on AWS S3 with CloudFront for fast global content delivery. I integrated Redis for caching and session management, and BullMQ for managing async job queues such as auction event processing and scheduled notifications. AWS S3 was also used for storing vehicle images and media assets. SendGrid and Twilio were integrated for automated email and SMS notifications at key auction lifecycle events, ensuring seamless communication throughout the bidding process.

I architected and led development of MedFlow, a high-scale healthcare management and medical billing ERP serving multi-department clinical workflows. I built a responsive React front end deployed via AWS Amplify, backed by a serverless Node.js and Express architecture on AWS Lambda for scalable computing. The system uses a hybrid database strategy — DynamoDB for fast patient record lookups and PostgreSQL for relational billing data, with GraphQL as the API layer. I implemented an automated PDF extraction pipeline where incoming emails with patient record attachments were auto-downloaded, processed through AWS Textract, and mapped into MedFlow's internal data structures, eliminating manual data entry. AWS S3 handled secure document storage with HIPAA-aligned access controls, and the platform supported HL7-compatible data formatting for interoperability with external healthcare systems. I established a GitHub Actions CI/CD pipeline, ensuring zero-downtime deployments for both front-end and back-end services.

I developed a high-performance front end for a video discovery platform that aggregates top trending videos from various websites across different categories. I built a sleek and modern user interface using React and Material UI, focusing on a seamless video browsing experience and intuitive navigation. I implemented advanced filtering and categorization features, enabling users to explore content by their interests. Collaborating with back-end developers, I integrated complex Python-based APIs for real-time data fetching and content synchronization, ensuring a fully responsive design across all devices and optimizing media player and grid layouts for mobile and desktop users.

I architected Smart Inventory, a production-ready multi-tenant SaaS platform to automate core retail operations. Built on the PERN stack with TypeScript, the front end uses React 19, Vite, TanStack Query v5, and Shadcn/UI, while the back end is a Node.js and Express API with Prisma ORM on PostgreSQL via Supabase. Core modules include real-time inventory tracking, automated sales and invoice generation, an HR suite with attendance and payroll, and granular role-based access control for multi-user authorization. I implemented a real-time POS supporting concurrent multi-terminal writes with FBR POS integration, using an async queue for instant receipt generation while submitting to FBR in the background. The platform features a voice-controlled POS interface supporting Urdu commands via Web Speech API with custom fuzzy product matching. State management is handled via Zustand with Zod for schema validation across all API boundaries. The system supports both cloud SaaS and on-premises installation, with the back end compiled to a binary using PKG for source code protection and AES-256-CBC-encrypted license files with machine fingerprinting.

At Spark AI, I worked on POWR.io, a suite of embeddable widgets used by thousands of small businesses on Shopify, Wix, and WordPress. My main focus was building and maintaining the React and Redux front end — lead-capture forms, checkout boosters, and social feeds. I used Styled Components throughout to keep theming flexible across 60+ plugins, and spent a lot of time optimizing bundle sizes so our widgets wouldn't hurt customers' PageSpeed scores.

At Spark AI, I worked on Ethera, a directory platform for mental health professionals. The UI had to feel calm and trustworthy, so we built the interface from scratch with custom CSS rather than a heavy framework. I built the therapist search and filtering system in React — users could narrow results by specialty, location, and availability. It was the kind of front-end work where the details actually mattered for real users finding help.

I designed and built Dubbing Studio end-to-end: a full-stack platform that automatically dubs short English videos into Urdu. I owned architecture, the FastAPI back end, ARQ background workers, and the React dashboard. Users register, upload clips, and track jobs while dubbing runs asynchronously, so the API stays responsive.

Stack: Python 3.11, FastAPI, MongoDB (Motor), Redis (cache and ARQ queue), JWT/bcrypt auth, Docker Compose (API, worker, Mongo, and Redis). Videos are owner-scoped, with multipart uploads and authenticated downloads.

Worker pipeline: FFmpeg extracts audio; OpenAI Whisper transcribes English; Groq (Llama 3.3 70B) translates to Urdu; Microsoft Edge TTS generates Urdu speech; FFmpeg merges audio into the final MP4. Job states: queued, processing, completed, failed.

Front end: React 19, TypeScript, Vite, Tailwind CSS 4—upload-and-dub, live status refresh, re-queue, and download. MVP pair: English to Urdu. AWS S3 and OpenAPI are configured for production scale.

I designed and developed the official marketing website for Draftly, my own SaaS company that builds intelligent, scalable enterprise software products. The site is built with Next.js and TypeScript for server-side rendering and optimal SEO performance, styled with Tailwind CSS v4 for a modern dark-themed design system. I integrated GSAP for high-performance scroll-triggered animations and Lenis for smooth scroll behavior, creating a polished, interactive user experience. Three.js was used to render the animated 3D geometric visuals on the hero section, adding a premium visual identity to the brand. The site is deployed on AWS S3 with CloudFront as the CDN for fast global delivery. The website showcases Draftly's product suite, including Draftly ERP — a live multi-tenant SaaS platform — and Draftly LMS, an AI-powered learning management system currently in development. The project demonstrates my ability to independently own a product from brand identity and UI design through to cloud infrastructure and deployment.