A little bit of a background about me, I accomplished my Bachelor’s diploma in Robotics and Automation at a high non-public college in India. The varsity ranks 63rd nationwide and high 10 (out of ~500 universities) within the state. I took up coursework on Machine Studying and AI for Robotics, which sparked my curiosity on this area. Whereas my time there formed my educational success, I typically questioned why I hadn’t thought of the trail of analysis at that time. I’ve come throughout lucky people who found their ardour for analysis throughout undergrad, and actively engaged with professors, collaborating on works for peer-reviewed journals or convention proceedings. It’s evident how publishing early on could be a game-changer, and I’ve witnessed the way it has positively formed the profiles of others within the area.
After my undergrad, I went on to work as a Machine Studying Engineer (MLE) at a startup. Though my time there was comparatively quick, I gained helpful expertise working with a well-liked deep-learning framework — TensorFlow. I contributed to a mission targeted on the quantization of neural networks in the course of the phases of coaching, analysis, and inference. By way of this course of, I discovered assemble neural community architectures, modify graph layers, and create kernels from the bottom up, permitting me to achieve a complete understanding of the underlying supply code of TensorFlow. Engaged on tasks (even private tasks) or roles like these will develop a robust sense of basis of deep studying optimization strategies and improve your considering functionality as a researcher. It’ll additionally present the boldness to not solely be the top consumer of a pre-trained algorithm however design customized or particular neural community fashions to fit your analysis aims.
Though some skip this technique of working within the trade (which is completely advantageous!), I personally noticed the worth in collaborating and gaining hands-on expertise that gave me an edge when persevering with to pursue my graduate research. This course of taught me successfully collaborate with teammates, drive tasks from conception to completion, and fearlessly deep dive to make and break issues within the course of. Total, my quick stint as an MLE was a really fulfilling journey and laid the muse for the subsequent step of my journey.
Irrespective of the place you might be in your journey, don’t fear concerning the missed alternatives. Take into consideration what you may management within the coming days, and put your effort in the direction of the subsequent most vital factor.
As a subsequent step, I progressed into enrolling in a Grasp’s program aiming to delve deeper into AI. Throughout this time, I loved taking the additional leap in constructing a portfolio web site, LinkedIn profile and tailoring my resume to showcase my skills. Principally, construct knowledgeable model. Have a look at your self two years from now and preserve including issues that make sense for future objectives. However that’s not all! This time additionally formed my communication abilities by numerous displays, successfully permitting me to convey my concepts. Collaborating with professionals from various backgrounds helped me construct a robust community that will final a lifetime. I additionally had a number of alternatives to work on thrilling hands-on tasks as a part of the coursework, which helped me construct a analysis portfolio and instilled a ardour for analysis deeply.
I consider that zeal for analysis is the curiosity about analysis’s potential to push the boundaries of a distinct segment area whereas nonetheless making a constructive impression locally. Moreover, you get alternatives to collaborate with fellow researchers throughout universities — this teaches you the method of getting ready for analysis lab conferences, important considering, and scientific writing. It additionally opens the door to attending technical conferences the place one can interact with consultants from various domains throughout the globe. This, mixed with the liberty to discover an space of curiosity, is what I name a perfect “Researcher’s World” — one the place curiosity, resilience, collaboration, and impression intertwine. My analysis experiences fall underneath the bigger umbrella of Accountable AI, and I aspire to be an skilled in privacy-preserving AI/ML programs. What intellectually stimulates me daily is the method of studying present literature, formulating analysis questions, and designing experiments to check my hypotheses.
In the course of the second 12 months of my Grasp’s program, there have been three pathways for me: coursework, project-based, and thesis-based. Proper from the beginning, I made the choice to pursue a thesis, because it ensured that the trail to a PhD would at all times stay open to me. The thesis is probably the most difficult path to Grasp’s completion and is similar to PhD dissertation however on a a lot smaller scale. I take into account myself lucky to have discovered a thesis advisor whose pursuits aligned intently with mine. I had gained curiosity in working with laptop imaginative and prescient primarily based generative fashions earlier than the Generative AI buzz erupted within the trade. After months of immersing myself within the related literature, a number of iterations of fastidiously designed experiments, and 1:1 conferences with my advisor, I efficiently accomplished my thesis on “Phoenix — A Federated Generative Diffusion Mannequin.”
Listed here are some key classes I discovered throughout this course of (1) Keep related by maintaining monitor of pre-prints and peer-reviewed works inside your area of curiosity (2) Be proactive in figuring out alternatives and addressing gaps within the dynamic area of AI (3) Put together for rigorous experimentation to check your method totally. For all Grasp’s college students aiming to pursue a PhD, right here’s my two cents: benefit from the journey of studying, and keep in mind to grab each alternative that goes past the curriculum.