Asad Ahmed

Being a sophomore student is not much time to gain much experience but having spent each and every time of that period I can say that I've learnt a few things and Machine Learning is among the top of my list. I've gone through the basic checklist of what everyone "has" to do for machine learning but what differentiates me is I've gone furthur and started applying it and developing new frameworks. Currently I've two research papers on Audio Style transfer and Accent Transfer under review and I'm involved in applying Reinforcement Learning based stabalizations on quadcopters which will be able to automate the drone travel under heavy weather conditions. Apart from that I'm good in IOT and Transmission sensors. My passion towards machine learning gives me a chance to give them autonomy.
The thrist of learning complex application of maths and computer science drives me to explore some of the unconventional paths. But apart from just solving complex issues I feel the need to help others to also learn things that might be complex. Making my mind occupied with complex studies is a coping mechanism of my depression but it doesn't pull me down. It motivates me to go further deep.
I've done few projects the most recent one that I was able to solve was involving "Accent Transfer", As the name suggest everyone has a charecterstic different accent or way of speaking the same sentences, to see in a more geographical area based viewpoint, British Accent and American Accent differ in certain "stresses" on words and pauses that make them sound different. Similarly when an arab person speaks English words there is a roll of tongue on the letter "r". This observation is easy to analyze when listened carefully by a trained person. But in order to program it and change ones accent to another was a tough task, that we encountered. Even though it sounds simple but in terms of machine learning what defines an accent is not explored properly, that made this task a problem because accent has to defined in terms of features. So we created a network that could learn charecterstic differences among different accent by using different language as data and then applied a general framework for changing the accent on a normal speech data, retaining the major part of speech yet able to change the way it sounds.
It gave me an insight into how diverse our world is and how differently humans evolves. But in terms of technical point if view extracting really low level features from data was a new thing to learn.
One of the projects I'm currently doing is using drones and reinforcement reinforcement learning which is a tough task from mechanical point of view because the drone has to be aerodynamically designed in order to reach a stable point. So we are trying to correlate the Reinforcement learning based training and physical parameters responsible, to design the mechanical structure of the quadcopter.
I've learnt a lot of physics and Mechanical Engineering aspect on this problem that was outside my comfort zone but yet, increased my spectrum to work on.
I'm a motivated, dedicated individual who thrives to learn things even from failures, I've been a constant patient of clinical depression but this ML/AI or Engineering in general is the reason I'm alive and it shouldn't be the reason to consider me weak, it is what makes me strong. Suffered from much in a short time but I've learnt a lot and wish to explore more.
Hackathons gives me a platform on national level to meet with people, fellow engineers with diverse background and learn a lot.
I might not know much but I've a constant zeal to learn more.