Xpromo’s website used to feel like a giant warehouse where everything was scattered—users couldn’t tell if items were local stock, factory-direct, or if they’d arrive before next Christmas. I jumped in as a full stack developer, wielding Next.js, React, and Tailwind CSS like a digital Swiss Army knife to reorganise the UX, add smart filters, and slap bold “Get your free quote” buttons so customers actually knew what was what.

The quote and mock-up process used to be a slow shuffle, so I built an automated workflow with instant mock-ups, approval steps, and notifications, keeping orders moving faster than a coffee-fuelled intern. I resolved all technical issues in accordance with the customer’s requirements—for example, I implemented real-time inventory visibility, optimised server-side rendering for speedy load times, and ensured the site looked gorgeous on any device.

Result: faster quotes, clearer UX, improved conversions and zero frustration-induced hair-pulling. Tech & impact: Next.js, React, Tailwind CSS, server-side rendering optimisations, real-time inventory visibility and instant mock-up generation.

Pacific Global Health’s website was like a sprawling library with books scattered everywhere—users struggled to find programs, resources, and key information quickly. I jumped in as a full stack developer, using MongoDB, Express, React, and Node.js (MERN) to reorganise the site, improve search, and create intuitive navigation so visitors could find exactly what they needed without wandering in digital limbo.

The content management and data workflows were a slow shuffle, so I built automated dashboards, dynamic filtering, and API integrations to streamline program listings, research resources, and partnership info. I resolved all technical issues in accordance with the customer’s requirements—for example, I optimised database queries for faster results, implemented server-side rendering for React pages, and ensured the site looked perfect on devices from phones to massive monitors.

Result: a smooth, responsive, and engaging experience that turns confused visitors into informed users while keeping everything fast, clear, and just a little bit charming.

Health.io was like a high-tech wellness playground, but users were getting lost in complex health data and slow responses. I stepped in as a full stack developer, wielding Next.js, Express, Prisma, HonoJs, MongoDB, Appwrite, Redis, Vector DB, AI tools like Llama33, and RAG workflows, along with Nodemailer for notifications, to build a fast, intuitive, and interactive experience.

The main challenge was making all the data—user metrics, AI insights, and recommendations—flow smoothly and be instantly accessible. I resolved all technical issues by integrating vector search for fast AI-driven insights, caching with Redis for speed, and secure backend data handling. I also made the UI responsive and engaging with real-time dashboards, dark mode, and seamless email notifications to keep users informed and motivated.

Result: the app turns complex health data into a fun, usable, and smart wellness companion—fast, reliable, and just charming enough to make users smile.

I wrestled with a bunch of generative AI models, trying to figure out which ones actually made sense and which ones were just making stuff up. To tackle that, I ran careful evaluations and measured every quirky output to ensure accuracy and efficiency. When models were being stubborn, I jumped in with some fine-tuning magic and hyperparameter tweaks to get them to behave like proper AI citizens.

Deploying them on SageMaker endpoints was like setting up a tiny AI café—real-time, ready for testing, and fully operational. Managing S3 datasets and artifacts kept me on my toes, ensuring everything was preprocessed, versioned, and securely locked down with AWS IAM best practices. I also played the role of cost-control hero, cleaning up unused instances and optimising resources so AWS bills didn't go rogue.

Result: the models were generating better outputs, the pipeline was humming smoothly, and I had proof that Python, AWS, and I make a pretty unbeatable team.