Hi, I'm Manoj Kumal.
A
Self-motivated, fast to learn, and enthusiastic programmer with an inquisitive mind that appreciates tackling complicated and difficult real-world challenges..
About
I’m currently a PhD student at Michigan Technological University and a Research Assistant working on agentic AI, multi-agent systems, and AI security. Previously, I completed my Bachelor’s in Computer Science and Information Technology from Tribhuvan University (Bhaktapur Multiple Campus). I bring over four years of hands-on experience developing AI-powered products, particularly in machine learning, computer vision, natural language processing and generative AI. My current interests lie in multimodal AI and advancing collaborative intelligent systems. Outside of research, I enjoy hiking and writing poetry. If our interests align, I would be happy to connect and exchange ideas.
- Languages: Python, C, C++
- Databases: MySQL, MongoDB
- Libraries: NumPy, Pandas, Matplotlib, Seaborn, OpenCV
- Frameworks: Flask, Keras, TensorFlow, Pytorch, Mlflow
- Tools & Technologies: Git, Docker, AWS, JIRA, Apache Airflow
Looking for an opportunity to work in a challenging position combining my skills in Software Engineering-Machine Learning, which provides professional development, interesting experiences and personal growth.
Experience
- Developed an Multi Agent Retrieval-Augmented Generation (RAG) chatbot tailored for the construction industry, improving operational efficiency and accuracy; deployed to over 20 clients at Krane.
- Fine-tuned domain-specific models for table detection and structure recognition, significantly enhancing the accuracy of document understanding in construction datasets..
- Optimized embedding models for improved semantic search performance; deployed scalable solutions using Kubeflow and KServe for robust, production-grade inference pipelines.
- Tools: Python, RAG, kubeflow
- Developed an Multi Agent Retrieval-Augmented Generation (RAG) chatbot tailored for the construction industry, improving operational efficiency and accuracy; deployed to over 20 clients at Krane.
- Fine-tuned domain-specific models for table detection and structure recognition, significantly enhancing the accuracy of document understanding in construction datasets..
- Optimized embedding models for improved semantic search performance; deployed scalable solutions using Kubeflow and KServe for robust, production-grade inference pipelines.
- Tools: Python, RAG, kubeflow
- Worked on an information extraction system for sales invoices and purchase orders using two approaches, the first of which was a purely rule-based approach.
- The second approach involved extracting header and footer information using LayoutLM, and extracting table information using Table Transformer.
- Fine-tuned LayoutLM and Table Transformer models on my own dataset and deployed model using Azure App service.
- Worked on Contrastive Language–Image Pre-training model to detect similar images using both images and corresponding text for training.
- Tools: Python, Flask, Opencv
- Developed and managed the information extraction software from citizenship using Flask, NER, OCR, and AWS.
- Worked on Optical Character Recognition and I have trained Custom Name Entity Recognition Model For Infromation Extraction From Nepali Citizenship.
- Performed Different Image Enhancement Techniques Like Morphological Operation, Filtering Techniques etcs and Image Alignment.
- Develop a Flask Web App Using Using AWS EC2 .
- Tools: Python, Flask, Opencv
- Learn About Python, Numpy, Pandas, Matplotlib
- Understand and Implement Machine Learning algorithms
- Learn Flask ,Mongodb and Postman method
- Study Deep Learning.
Projects
Agetic AI chatbot for Construction Domain
- Tools:Python, Langgraph, Numpy, Kubeflow, Git
- chatbot
Intelligent Information Extraction System
- Tools:Python, Transformer, Numpy, Git
- chatbot
Skills
Languages and Databases
Python
HTML5
CSS3
MySQL
Libraries
NumPy
Pandas
OpenCV
scikit-learn
matplotlib
Frameworks
Flask
Keras
TensorFlow
PyTorch
Other
Git
AWS
Heroku
Linux
Education
Michigan Technological University
Houghton, Michigan, United States
Degree: Doctor of Philosophy in Computational Science and Engineering
Percentage: Inprogress
Institute of Science and Technology, Tribhuvan University
Kathamandu, Nepal
Degree: Bachelor of Computer Science and Information Technology
Percentage: 80%
- Data Structures and Algorithms
- Database Management Systems
- Operating Systems
- Machine Learning
- Computer Vision
Relevant Courseworks: