Prakalp Srivastava
Founder at Eigenwave (Stealth)
Mountain View, California, United States
Summary
🤩
Rockstar🎓
Top SchoolPrakalp Srivastava is a founder and seasoned compiler engineer with 12 years of experience building ML infrastructure and production compilers for CPUs, GPUs, and accelerators. He has held engineering and research roles at Google Brain, OctoML, and Netflix, and now leads a stealth startup from Mountain View focused on advancing ML systems. His open-source work on high-profile projects like Apache TVM and OpenXLA demonstrates deep expertise in layout transformations, constant folding, codegen optimization, and numerical dtype handling—practical contributions that reduce compile time and improve correctness for modern accelerators. Holding a PhD from UIUC and an MTech/BTech from IIT Kanpur, he blends rigorous academic training with product-focused engineering and a track record of shipping backend compiler features that matter in real-world ML workloads.
12 years of coding experience
14 years of employment as a software developer
Indian Institute of Technology KanpurDoctor of Philosophy - PhD Computer Science, Doctor of Philosophy - PhD Computer Science at University of Illinois Urbana-Champaign
Github Skills (22)
c-language10
operation10
machine-learning10
compiler-compiler10
hla10
xla10
compiler10
cprogramming-language10
numerical9
computation9
python9
deeplearning-ai9
deep-learning9
tensor9
numerics9
Programming languages (3)
Github contributions (5)
apache/tvm
Feb 2023 - Jan 2024
Open deep learning compiler stack for cpu, gpu and specialized accelerators
Role in this project:
Back-end Developer & Compiler Engineer
Back-end Developer & Compiler EngineerContributions:42 reviews, 11 PRs, 59 comments in 11 months
Contributions summary:Prakalp primarily focused on enhancing the TVM compiler stack, with contributions spanning multiple areas. They addressed issues in the TOPI (Tensor Operator Inventory) module, specifically fixing pooling operations to return errors for tiled layouts. They also implemented layout transformation analysis at the PrimFunc level, enabling optimization of data layouts within the compiler, and worked on constant folding, adding a pass to evaluate and fold Relax operations. The user further made improvements to the Cutlass codegen, optimizing compilation time, and addressed a bug related to bool datatypes in the block memory planning process.
metalvulkancompilertensoropencl
openxla/xla
Oct 2019 - Mar 2021
A machine learning compiler for GPUs, CPUs, and ML accelerators
Role in this project:Back-end Developer & ML Engineer
Back-end Developer & ML EngineerContributions:5 commits in 1 year 5 months
Contributions summary:Prakalp primarily contributes to the XLA compiler, focusing on backend and machine learning-related tasks. They removed obsolete code and added a builder for HLO Copy instruction, enhancing the compiler's functionality. Furthermore, the user worked on converting data types (F64 to BF16), indicating involvement in numerical computation and optimization, likely for hardware acceleration. Finally, they fixed a bug related to instruction ordering, crucial for the compiler's correctness.
compilercommunity-drivenmachine-learningmodular
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.