I build autonomous infrastructure to help solve high-stake technical bottlenecks using agentic workflows and data pipelines.
A client wanted to build an e-commerce store for their walk-in pharmacy, but they didn’t have a catalogue complete with detailed-descriptions, product images, and product costs.
I designed and deployed a resilient scraper built on Beautiful Soup. The aim was to have a complete catalogue of over 2,000 SKUs, where the images needed to be of a specific size and resolution.
Although the scraping was successful. A challenge arose, in that the images were not only in different formats, but also visibly low-resolution and uniform. Besides, processing these images would have taken over 12 hours on my simple CPU based machine.
I solved this by finding free tier GPUs on Google Colab, chaining the data to Real-ERSGAN tool and specifying the desired outcome, where the results would automatically be output to my Google Drive upon completion.
The processing took just under 45 mins, with satisfying results in the end by all standards.
Portfolio: [https://dev.sephan.co.ke\]
Rate: $55/hr or project based.
If you have similar or related projects. PM me.