Hi, I'm Richard Khewa Limbu
Computer Engineering Graduate & AI Developer
Building intelligent workflows, advanced RAG architectures, and robust backend systems.
About Me.
I am a focused and driven Computer Engineering graduate specializing in Machine Learning, Deep Learning, and backend ecosystem architectures. I engineer intelligent software solutions by connecting structured python frameworks with advanced generative models and robust vector pipelines.
With a strong academic foundation from Khwopa College of Engineering, I specialize in Generative AI and Machine Learning systems, focusing on LLM optimization, retrieval-augmented and graph-based agentic workflows, and production-grade backend development using FastAPI and Django for scalable RESTful and AI-driven applications.
- AI & Machine Learning: PyTorch, Scikit-learn, LangChain, LangGraph, Pandas, NumPy
- Backend Development: Django, Django REST Framework, FastAPI
- Databases & Environments: PostgreSQL, MySQL, MongoDB, Docker
- Tools & Platforms: Git, GitHub, VS Code, Jupyter Notebook, Linux
Django Intern — Sajilo Life Pvt. Ltd.
Engineered responsive web interfaces using Django MVT structure, integrated secure multi-provider user authentication layers using Django Allauth, and handled complex data queries with Django ORM.
Bachelor of Computer Engineering
Khwopa College of Engineering
Higher Secondary Education (Science)
Takshashila Academy and College
Core Expertise.
Generative AI & RAG
Designing stateful multi-agent systems using computational graphs, contextual history-aware query expansion, and semantic chunking storage retrieval paths.
Deep Learning & Vision
Training and evaluating deep architectures, treating extreme class distributions using custom samplers, and implementing generative modeling architectures.
Backend Architecture
Building high-throughput, structured web backends and microservices featuring rigid data schemas and clean authentication layers.
My Works.
Multi-AI Agent Workspace
An advanced system orchestrating specialized LLM agents featuring automated task delegation, runtime tool invocation hooks, and stateful conditional routing graphs.
Contextual RAG Engine
Document analysis ecosystem integrating history-aware prompt formatting, similarity based document splitting, and vector lookup embeddings.
Skin Lesion Classification Model
Fine-tuned ResNet18 convolutional architectures on HAM10000 datasets to classify 7 lesion phenotypes, scoring 84.2% validation accuracy using custom augmentation routines.
CycleGAN Manga Colorizer
Designed and implemented a grayscale manga colorization system using CycleGAN, in a team of four. Constructed a CNN classifier and an optimized training pipeline to enhance model performance.
AI Anime & Manga Recommender
Anime and manga recommendation system powered by RAG over 85,000+ titles. Combines FAISS vector search with MMR retrieval, LLM-driven query expansion, and conversational memory with real-time streaming.