System Architect // Computer Engineering

Hi, I'm
Yamraj Khadka.

A Computer Engineer exploring the future of Artificial Intelligence. I enjoy building AI systems, conducting research, and developing applications using machine learning, deep learning, computer vision, and large language models.

SYSTEM REPOS MODEL HUB
Diagnostic — Kathmandu, NPL

$ whoami

Yamraj Khadka | AI Researcher

$ location

Kathmandu, Nepal 🇳🇵

$ impact_adoption

["Model_Adopted_24h", "CPU_Hardware_Ready", "4_Live_Apps"]

$ status

Building impactful technology...

$ _

0.674

U-Net mIoU

13.5GB

Mistral FP16

24h

Community Adoption

100+

Students Mentored

Journal Publication · JoEIS Vol 5

"Land Cover Segmentation from Satellite Imagery Using U-Net with Custom Loss"

My research engineered a custom Deep Learning framework for automated mapping, optimizing land expansion analysis and disaster response for Nepal's unique topography.

View Publication Codebase

📈 Impact & Adoption

Community Re-quantization

Adopted within 24h by mradermacher

Format Diversity

Available in FP16, GGUF Q4, Community Q2-Q8

CPU Accessibility

4GB model runs on consumer hardware CPUs

Deployments

4 live applications + 2 production APIs

U-Net Segmentation

System Manifest

/active_technical_nodes

e-wakil: Legal LLM

Fine-tuned Mistral-7B on Nepal's Penal Code. Pipeline includes PDF extraction, RAG retrieval, and GGUF quantization.

Mistral-7B RAG

Agentic Nexus

5-agent orchestration system using LangGraph. Features autonomous memory loops and content safety modules.

LangGraph

Production MLOps

Reproducible ML pipeline for trip prediction. Integrated MLflow for tracking and Optuna for optimization.

MLflow

Engineering Log

Deep Learning build series

Read All Posts →
Series: Day 17

YOLO & Semantic Segmentation

Deep dive into real-time detection and pixel-level segmenting for agricultural tracking.

Vision_Core
Series: Day 14

Transfer Learning Mastery

Optimizing VGG16 weights for CIFAR-10 with 73.13% accuracy with minimal compute.

Transfer_Learning
Series: Day 11

ML Project Diagnostics

Implementing Andrew Ng's strategy for diagnosing system errors and handling mismatched sets.

MLOps_Strategy