baudhya - Overview

Hi 👋, I'm Siddharth Baudh

Scientist at DRDO India | Ex-IISc | Systems & ML Infra Enthusiast

🎓 M.Tech in Computer Science, Indian Institute of Science (IISc), Bangalore
🔬 Scientist, Defence Research and Development Organisation (DRDO), India
💼 Former Software Engineer @ Finoit Technologies
📫 Reach me: siddharthbaudhya6216@gmail.com | LinkedIn

🛠️ Interests & Specialties

• Systems Programming: eBPF, Kernel Internals, OS-Level Instrumentation
• Compilers & Security: LLVM FunctionPass, syscall tracing, sandboxing
• Machine Learning Infra: CUDA, AVX512, parallel training, PyTorch, Ring-AllReduce
• High-Performance Systems: Branch predictors, simulation (ChampSim), memory checkpointing
• Languages & Tools: C++, Python3, eBPF, CUDA, Docker, Git, FastAPI, Celery

🚀 Highlight Projects

• 🧩 Memory Checkpointing with eBPF – Checkpoints heap memory at runtime using syscall-level instrumentation
• 🔐 In-Kernel Sandboxing – Combines LLVM IR instrumentation + eBPF for enforcing syscall-level policies
• ⚙️ Optimizing Dilated Convolution – CUDA + SIMD (AVX2/AVX-512) version achieving up to 800x speedup
• 📈 Data-Parallel LeNet – Distributed training using Ring-AllReduce; 50% faster with same accuracy

🧠 Background

I'm passionate about building robust, high-performance systems at the intersection of systems software, machine learning infrastructure, and computer security. I enjoy working close to the OS and hardware, optimizing performance, and developing infrastructure for future AI systems.

📌 Let’s Connect

Open to collaborations, research projects, and deep tech discussions in OS, ML systems, compilers, and secure computing.

📬 Email: siddharthbaudhya6216@gmail.com
🔗 LinkedIn: linkedin.com/in/baudhya