Prashant Aryal

AI/ML Engineer

Kathmandu, Nepal

About Me

I'm an Associate AI/ML Engineer currently working at Sajilo Ventures Pvt. Ltd., with a Bachelor's degree in Computer Science and Information Technology from St. Xavier’s College, Tribhuvan University. I have a strong foundation in computer science, with hands-on experience in machine learning, deep learning, and large language models to build practical, AI-powered solutions. I've worked on projects ranging from plant disease detection to AI chatbots, and have co-authored a research paper on corn leaf disease diagnosis. I'm always eager to keep learning and growing in the fields of Data Science and AI. Outside of work, you'll often find me playing football, cricket, or table tennis, exploring new places, or simply enjoying music and nature.

Education

Bachelor of Science in Computer Science and Information Technology 2019 - 2024

St. Xavier's College, Tribhuvan University, Kathmandu, Nepal

Major courses: Database Management System, Artificial Intelligence, Design, Analysis and Algorithm, Data Warehousing & Data Mining, Data Structure and Algorithm

Higher Secondary School (Grade XI & XII) 2017 - 2019

National School of Sciences (NIST), Lainchour, Kathmandu, Nepal

Major courses: Physics, Chemistry, Mathematics, Computer Science

Work Experience

Associate AI/ML Engineer at Sajilo Ventures Pvt. Ltd. March 2025 – Present

Developing a real estate chatbot for a Dubai-based client using FastAPI, MongoDB, and FAISS. Utilizing Alibaba-NLP embeddings from HuggingFace to encode property data and Llama 3.3 70B LLM through Groq's API to process user queries in natural language, delivering real-time, accurate responses for property-related inquiries.
Researched and implemented saliency-based object detection and image augmentation using OpenCV and TensorFlow, applying bounding box expansion, image cropping, and data augmentation techniques (rescaling, random rotations, zooms, shifts, flips, and brightness adjustments) to enhance deep learning models for anti-poaching monitoring.

Data Science Intern (Virtual) at CodSoft November 2024 – December 2024

Developed supervised learning models for classification, regression tasks and predictive analytics in applied data science projects using Python, Scikit-learn, NumPy, and Pandas.
Performed data preprocessing, feature engineering, and model evaluation techniques to solve real-world data problems, with a strong focus on practical implementation.

Software Intern at LIS Nepal Pvt. Ltd March 2024 – June 2024

Worked on data integration projects using Workato, focusing on retail operations such as order management systems (OMS) and transfer processes for Bluemercury, a US-based luxury cosmetics retailer.
Gained practical experience in Python, SQL, Data Warehousing, and ETL (Extract, Transform, Load) processes.
Contributed to testing processes, prepared UTDs (User Test Documents), and enhanced quality assurance efforts.

Projects

BudgetGPT: Nepali Fiscal Policy Chatbot

Developed an AI chatbot to answer queries related to Nepal's national budget, offering fast access to fiscal data from the official Nepal Government Budget speech (English and Nepali version) through semantic search powered by FAISS.
Integrated LangChain, HuggingFace embeddings (BAAI/bge-m3 for English and All-MiniLM-L6-v2-Nepali for Nepali), CTransformers, LlamaCpp and FAISS to enable efficient large language model (LLM) inference and accurate bilingual text generation across both Mistral-7B (for English) and Nepali-LLM-Instruct (for Nepali).
Deployed a Flask-based web application with a responsive UI, enabling multiple user queries in a single session and maintaining in-memory chat history.

Technologies: Python, LangChain, HuggingFace, FAISS, CTransformers, LlamaCpp, Mistral-7B (GGUF), Nepali-LLM-Instruct (GGUF), Flask, langdetect, PyPDF2

SmartAgro: Plant Leaf Disease Classification System

Developed a Plant Disease Classification system using Convolutional Neural Network (CNN) architectures such as ResNet-50V2 and VGG-19 to classify diseases affecting plant leaves, demonstrating their effectiveness in agricultural image classification.
Integrated the ResNet-50V2 model into a Flask-based web application, offering farmers and agricultural experts a user-friendly interface to access and utilize the system for disease diagnosis and management.

Technologies: CNN, TensorFlow, Transfer Learning (ResNet-50, VGG-19), Google Colab, Flask

Email Spam Classifier using Machine Learning

Applied machine learning algorithms to tackle spam classification problems, comparing their performance to gain deeper insights into the issue.
Designed a web application using Flask, leveraging the Multinomial Naive Bayes Algorithm for its high accuracy in effectively filtering email spams.

Diabetes Prediction using Machine Learning

Utilized the K-Nearest Neighbors Classifier to develop a model that incorporates various parameters such as glucose levels, blood pressure, skin thickness, insulin levels, and body mass index to predict whether an individual has diabetes.

Prashant Aryal

About Me

Education

Work Experience

Projects

Publications

Get In Touch