C++ developer · ML infrastructure engineer.
Currently at Research Commons — building a C++ AI inference engine from the ground up, writing SIMD (AVX, NEON) and CUDA kernels, and shipping the Kubernetes-native distributed training platform (Python client + Go operator + Volcano).
- Systems C++ — memory, threading, execution pipelines, JIT, compiler IRs. Reading PyTorch and tinygrad internals for fun.
- ML infrastructure — distributed training orchestration, checkpointing, GKE provisioning, observability.
- Graphics and game engines — from-scratch ECS engines, OpenGL, Unreal, Unity. Personal projects are mostly here.
Posts on the Research Commons engineering blog:
- Single Node Training with GKE and Ray
- Functional-Gemma: fine-tuning for edge function calling
- Intro to Neural Networks
- PyTorch Overview — From Research to Prod
- Portfolio → shivanshahuja.dev
- Resume → shivanshahuja.dev/resume
- Projects → shivanshahuja.dev/projects
- Email → shivansh.ahuja91@gmail.com