Welcome

About Me

I am a PhD candidate in Computer Science at Purdue University, specializing in Scientific Computing, High Performance Computing, Machine Learning, and Deep Learning. I am an experienced full-stack software engineer with well-rounded design and development skills in highly distributed applications.

My passion lies in building high-performance computing systems and developing scalable solutions for complex computational problems. I consider research as a creative art and enjoy the challenge of taming complex systems to achieve optimal performance. My work focuses on parallel computing, hybrid programming models, and the intersection of HPC with modern machine learning techniques.

Previously, I held senior engineering positions at LinkedIn R&D, Electronic Arts, and Microsoft, where I worked on cutting-edge distributed systems, big data platforms, and cloud computing infrastructure. I have been recognized as an "Ambassador" by "Leaders for Good" for human-centric leadership practices and am enthusiastic about mentoring and building high-performance teams.

Research Interests

  • Parallel and Distributed Computing: Performance optimization of MPI-based multigrid solvers for scientific simulations.
  • Deep Causal Discovery: Exploring nonlinear and probabilistic causal inference methods for time-series data.
  • Hybrid GPU-Accelerated Architectures: Developing CUDA-parallelized models for large-scale agent-based and PDE systems.
Scientific Computing High Performance Computing Machine Learning Deep Learning Parallel Computing Distributed Systems Cloud Computing

Research Philosophy

I view research as both a creative and intellectual art—an iterative process of designing, building, and understanding intelligent systems. Sharing research through publications or web platforms is not merely documentation, but an act of showcasing creativity, curiosity, and rigor. I find joy in writing programs that evolve into systems capable of understanding intent—transforming abstract ideas into tangible intelligence. To me, that process embodies the essence of “taming the beast.”

My research interests lie at the intersection of distributed and parallel machine learning, with a focus on large-scale generative AI, algorithmic fault tolerance, and parallel causal inference. I explore how distributed learning frameworks can be made more resilient and efficient through intelligent recovery, redundancy-aware algorithms, and adaptive communication strategies. A central goal of my work is to make LLMs and distributed generative models more scalable, fault-tolerant, and self-correcting in heterogeneous computing environments.

I am also deeply interested in the systems perspective of AI—how algorithmic and architectural co-design can enable high-performance, reliable learning at scale. This includes research on hybrid and distributed programming models (MPI + CUDA, MPI + OpenMP, MPI + Pthreads), as well as leveraging GPUs and accelerators for efficient large-scale model training. The challenges of communication, synchronization, and recovery in distributed AI systems inspire my broader work in fault-tolerant distributed ML and parallel causal discovery.

Computing Perspective

In high-performance computing (HPC) and AI, progress is often measured in flops, speedups, and scaling efficiency—but I see it as a dialogue between algorithms and architecture. Parallel and distributed systems are living ecosystems where processes must cooperate, fail gracefully, and recover intelligently. My research seeks to understand and enhance this interaction—transforming computational resources into intelligent collaborators rather than passive executors.

Building a massively parallel system is an engineering triumph; writing algorithms that harness its full potential with elegance and resilience is the true form of worship in computational science.