> whoami

class SagarSahu:
    role      = "Senior Software Engineer"
    exp       = "10+ years shipping production systems"
    focus     = ["Heterogeneous Computing", "Edge AI", "Systems Programming", "Cloud & Security"]
    languages = ["Python", "C++", "C", "Rust", "Shell", "TypeScript"]
    currently = "Programming AMD GPUs & NPUs for accelerated AI inference"

I build software that runs close to the metal — from GPU kernels and NPU inference pipelines to cloud-scale security agents deployed on millions of endpoints. I care about performance, correctness, and writing code that other engineers enjoy reading.

---

🛠️ Tech Stack

Languages

AI / ML / Accelerated Computing

Cloud & Infrastructure

---

📌 Featured Projects

Project	What it does	Tech
gemem-AMD	9-session tutorial series: heterogeneous computing on AMD GPU (RDNA 3.5) + NPU (XDNA 2) — from OpenCL kernels to INT8 quantization	Python OpenCL ONNX Runtime PyTorch
YoutubeSummarizer	AI-powered YouTube video summarizer with periodic updates	Python AI/ML
sample-app-aoai-chatGPT	Web chat interface powered by Azure OpenAI ChatGPT	TypeScript Azure OpenAI
distributed_computing	Distributed systems fundamentals — consensus, replication, fault tolerance	C MPI

---

📊 GitHub Stats

  

---

🧠 What I'm Working On

• 🔥 Heterogeneous Computing — Writing GPU/NPU kernels for AMD Ryzen AI (RDNA 3.5 + XDNA 2)
• 🧮 Performance Engineering — Amdahl's Law → Roofline Model → Kernel Fusion → real benchmarks
• 🤖 Edge AI — INT8 quantization, ONNX Runtime execution providers, on-device inference
• ☁️ Cloud + AI — Azure OpenAI, Databricks, building production ML pipelines

---

🤝 Connect

---

_{⚡ "Make it work, make it right, make it fast." — Kent Beck}