Naidu Saikiran

My Educational Background and Research Interests

From the early stages of my undergraduate education, I developed a deep fascination with artificial intelligence. I pursued a BTech in Computer Science from an unreputed university, but despite the limitations of my academic environment, I sought to build a strong foundation in AI and machine learning. In my second semester, I began learning machine learning through NPTEL and Andrew Ng’s Stanford course, which sparked my curiosity about the theoretical and practical aspects of AI.

Over the years, I expanded my knowledge by diving into deep learning through NPTEL, completing David Silver’s Reinforcement Learning (RL) lectures, and progressing up to actor-critic methods in Sergey Levine’s RL course. I also attempted CS224n (Natural Language Processing with Deep Learning) and explored DSA and web development, though my primary interest remained in AI.

However, I faced significant challenges in securing research opportunities during my undergraduate years. Due to financial constraints, I was unable to pursue studies abroad immediately after graduation, so I turned to GATE preparation as a means to enter IITs/IISc for research. Unfortunately, despite dedicating two years to GATE, my performance did not meet my expectations, leading me to reconsider my path.

My Research Interests

My primary research interests lie in AI safety, mechanistic interpretability, and alignment. My concern about AI alignment deepened after reading a blog post on LessWrong, which led me to explore the field more seriously. I actively follow researchers like Jan Leike and Neel Nanda, whose work in mechanistic interpretability and AI safety inspires me.

I aspire to become a good researcher in mechanistic interpretability or AI alignment and contribute meaningfully to the field. To bridge my knowledge gaps, I have been following Redwood Research’s AI safety research curriculum and simultaneously working on improving my PyTorch skills for tensor manipulation and NLP model building. I also plan to complete CS224n, including the final project, within the next three months.

Given my strong mathematical background and programming skills, I believe I have the potential to make significant contributions to AI safety. My long-term goal is to pursue a PhD in AI safety and alignment at a top university like Stanford. Although I lack prior research experience, I am determined to prove my capabilities through self-study, projects, and collaborations.

Despite the setbacks in my academic journey, my passion for research remains unwavering. I am actively seeking opportunities, whether in pre-doctoral positions at Microsoft, Google, or other research labs, that will allow me to gain valuable experience before applying for PhD programs. My journey has been unconventional, but I am committed to making an impact in the field of AI alignment and mechanistic interpretability.