Blog — Ishaan Shah

Notes on building software — open source, infrastructure, and side projects.

• Testing How LLM Inference Scales Across GPUsSep 22, 2026

  I benchmarked and profiled Llama 3.1 8B from one to four GPUs. Here is the mental model I wish I had before touching vLLM, Nsight, or a single tuning flag.

  12 min read
• My Overengineered Side-Project StackNov 28, 2025

  Why I moved my side projects off Vercel free tiers onto EC2 + Dokploy — and how self-hosting made me ship more, not less.

  6 min read
• My Journey through GSoC'23Oct 4, 2023

  A summary of my time as a Google Summer of Code contributor with Postman — building a scalable OpenAPI search over 80,000 GitHub files with ElasticSearch.

  8 min read