Engineering production-grade AI pipelines, RAG systems, deep learning diagnostics, and modular backend APIs.
ZENITH-RAI / rag-chatbot-platform
Modular enterprise RAG system featuring multi-stage document ingestion, dense-vector embeddings, and BM25 hybrid search. Fully containerized with Docker Compose across cache, API, and vector database nodes.
ZENITH-RAI / ecommerce_backeendprj
High-throughput backend REST API built with FastAPI, Pydantic v2, and SQLAlchemy 2.0. Includes JWT authentication, bcrypt password hashing, role-based access control (RBAC), Alembic migrations, and 90%+ PyTest coverage.
ZENITH-RAI / EYEDISEASEPREDICTION
Deep learning computer vision application utilizing Convolutional Neural Networks (CNNs) trained on 4,000+ fundus scans across four disease classes, achieving 96.2% test accuracy with OpenCV preprocessing and FastAPI inference.
ZENITH-RAI / final-year-project
Machine learning valuation system leveraging XGBoost and AdaBoost (R²: 0.87) to predict secondary market car prices, backed by K-Means segmentation and a FastAPI marketplace backend with eSewa digital wallet integration.
PRODUCTION CLINICAL DIAGNOSTIC
Clinical classification system utilizing supervised learning algorithms to assess patient parameters (blood pressure, cholesterol, ECG indicators) for rapid and reliable cardiovascular disease risk evaluation.
ZERO-CLOUD PRIVACY ARCHITECTURE
Generative AI application enabling zero-leakage enterprise document Q&A using local open-source LLMs via Ollama, LangChain chunking pipelines, and vector semantic similarity search for confidential documents.
Comprehensive stack engineered across machine learning, deep learning, generative AI, and resilient backend systems.
As you scroll, the cyber circuit timeline traces academic progression and milestone achievements.
Old Baneshwor, Kathmandu
Completed secondary schooling with distinction honors. Fostered rigorous foundational mastery across mathematics, computer fundamentals, and scientific inquiry.
Higher secondary academic coursework with an emphasis on Physical Sciences, Advanced Mathematics, and Computer Science. Built foundational programming competence and computational logic.
Kathmandu, Nepal
Affiliated to Tribhuvan University (TU)
Bijayachowk, Gaushala, Kathmandu
4-year Bachelor of Science in Computer Science and Information Technology. Rigorous university curriculum covering operating systems, compiler design, artificial intelligence, computer networks, and enterprise database systems.
Complete semester-wise academic curriculum from Tribhuvan University Institute of Science and Technology (IOST).
Hands-on production projects, model architecture prototyping, and full-stack software development.
Machine Learning & Backend Development
Kathmandu, Nepal
Supervised & Unsupervised Modeling: Engineered and tuned regression and classification models (multilinear regression, logistic regression, decision trees, XGBoost, and AdaBoost) leveraging systematic feature engineering and hyperparameter optimization.
Deep Learning Computer Vision: Architected CNN and ANN neural networks utilizing PyTorch and Keras with Adam and RMSProp optimizers, batch normalization, and dropout layers—achieving 96.2% test accuracy on retinal pathology image classification.
Generative AI RAG Chatbot: Developed a full-stack Retrieval-Augmented Generation (RAG) platform using LangChain, ChromaDB vector storage, and FastAPI, integrating K-Means and DBSCAN clustering for automated dataset partitioning.
Containerization & Backend Hardening: Packaged microservices with Docker, orchestrated PostgreSQL database schema migrations using Alembic, and enforced secure user authentication through JWT and bcrypt hashing.
Direct access to open source repositories, architecture blueprints, and source code implementations.
Enterprise modular e-commerce REST API built with FastAPI, SQLAlchemy 2.0, Pydantic v2, Alembic migrations, and role-based access control. Includes comprehensive PyTest suites.
git clone https://github.com/ZENITH-RAI/ecommerce_backeendprj.git
Deep learning retinal scan diagnostic classifier using Convolutional Neural Networks (CNNs). Trained across 4 ophthalmic pathology classes with real-time FastAPI inference.
git clone https://github.com/ZENITH-RAI/EYEDISEASEPREDICTION.git
Machine learning vehicle resell price prediction platform with 0.87 R² score, integrated with K-Means market segmentation, FastAPI backend, and eSewa payment gateway.
git clone https://github.com/ZENITH-RAI/final-year-project.git
Production RAG enterprise document intelligence platform with dense vector embeddings and BM25 hybrid search. Proprietary codebase. Live demonstration and architecture walkthroughs available upon request.
Actively open to software developer internships, junior backend/AI roles, and high-impact engineering collaborations.
Connect directly for internship discussions, technical interviews, or project inquiries.