Kunal Pathak

About Kunal Pathak

Compiler/Codegen Engineer specializing in backend optimization, register allocation, and ISA-centric performance. Currently at NVIDIA working on the LLVM backend, where I focus on register allocation improvements, instruction selection/codegen robustness, GPU architecture-specific optimizations, and low-level performance work across the toolchain.Previously at Microsoft, I spent 15+ years on the.NET Runtime JIT (RyuJIT) backend, owning phases such as register allocation, encoding, peephole and lower-level optimizations, loop alignment, TLS/struct return fixes, and ISA-targeted work across x64, Arm64, and Apple silicon. I also led ARM64/SVE initiatives, delivering new vector instructions, LSE atomics, load/store pattern improvements, and intrinsic support used in production today. Before RyuJIT, I worked on Chakra (IE/Edge JavaScript engine), implementing language features, Node.js on Chakra, and performance enhancements for AcmeAir/TechEmpower benchmarks. Earlier, I contributed to ML-guided program synthesis via Microsoft PROSE.I enjoy building optimizations that are visible at the assembly level, solving performance bugs that require register-level thinking, and bridging the gap between language semantics and hardware realities.MS in Computer Science (RIT), specializing in Programming Languages and Compiler Design.

Experience

1. Compiler Engineer

   NVIDIA

   Jul 2025 — Present · Redmond, WA, US

Education

• Rochester Institute of Technology

  Master of Science, Computer Science

  2007 — 2009

• Vishwakarma Institute Of Technology

  Bachelor of Engineering, Computer Science

  2000 — 2004

Skills

• Wcf
• Software Development
• Javaserver Pages (Jsp)
• Servlets
• Software Design
• C
• Programming
• Aspectj
• Wpa
• Xsl
• Xml
• Visual Studio
• Algorithms
• C++
• Xslt
• Javascript
• Jsp
• Sql
• .net Framework
• C#
• Oop
• Agile Methodologies
• Java
• Html
• Windbg
• Computer Science