Ibrar Yunus

Architecting Gifftid's Vision:

• Architected and implemented the complete GenAI platform, including GCP infrastructure, CI/CD pipelines, data processing/persistence pipelines, access controls, chatbots, and vector databases, enabling rapid AI-powered solutions for SMEs.

• Working hand-in-hand with the CEO, I provided the strategic vision that redefined Gifftid's impact. By pinpointing untapped opportunities and pioneering AI tools purpose-built to empower SMEs, I fueled remarkable company growth and firmly established Gifftid as the leading force in AI-driven SME solutions.

Prominent work:

Worked on a project to use Machine Learning techniques on a data-set of video segments to find statistical relations between the user preferences when encoding video and the outcome. Should we compress video game streaming differently to live phone streaming? What about different games? Can we change how we encode different sub-scenes of the same video? With video streaming being the dominant source of data transfer on the Internet, this is an important problem. This project followed on from a PhD project which looked at non-streaming video and images in an equivalent way, predicting encoding times, output file sizes and output video quality based on video features and encoding parameters. Regression tasks were performed using a data-set of 2500 Full-HD videos on a computer cluster of 12 nodes (each node being Intel Xeon 3.4GHz CPU running Scientifc Linux) using Intel AVX instructions.

This also involved research into the concept of dynamic resolution, which further enhances this goal by suggesting a new way to encode videos, where parameters are adjusted per segment.