KUSHAL RAJ SHARMA
Mississauga, ON, Canada
Professional Summary
AI Enthusiast with over 3 years of experience in Machine Learning, Natural Language Processing, Large Language Models, Data analysis, proficient in Python. I continuously challenge my knowledge and skills by competing in hackathons and completing online courses to enhance my foundation. I am seeking an opportunity where I can spur innovation and leave a lasting impact by applying my technical expertise and collaborative abilities.
Professional Experience
Machine Learning Engineer
Omdena
Remote
July 2023 - September 2023
- • Collected and preprocessed over 500,000 Nepali and English language data samples from diverse sources, including news articles, social media, and academic publications.
- • Produced summary statistics and visualized data distribution, identifying key anomalies led to a 15% improvement in data accuracy for model training.
- • Developed ML Model for Sentiment Analyzer and Entity Extraction in collaboration with team members.
- • Conducted sentiment analysis on 10,000+ data points, diminishing bias detection time by 10%.
AI Lead
Sherpa.Codes
Butwal, Nepal
September 2021 - May 2023
- • Led a team of 20+ annotators, ensuring quality control and timely completion of data annotation projects, resulting in creation of a high-quality labelled dataset operated for training machine learning models.
- • Collected data and preprocessed data using spacy, enabling efficient acquisition and preprocessing of large-scale data for NLP projects.
- • Collaborated with cross-functional teams to prioritize data needs for machine learning models, leading to improved model accuracy and relevance in real-world applications.
- • Utilized supercomputer resources to optimize efficiency and reduce computation time by 50% enabling faster experimentation and model iteration.
- • Mentored junior students through final year projects by applying Machine Learning techniques, offering hands-on guidance in model development, and data analysis, resulting in a 100% project success rate.
Artificial Intelligence Intern
National Innovation Center
Kathmandu, Nepal
May 2021 - August 2021
- • Implemented Convolutional Neural Networks (CNN) from scratch for image classification tasks, achieving an accuracy rate of 90% on a benchmark dataset.
- • Reviewed and analyzed research papers on Autonomous Self-Supervised Learning, leading to the identification of novel techniques and improved model performance by 15%.
- • Recreated the Plant Disease Detection Model using Pytorch, reducing inference time by 30% through model optimization techniques.
Skills
Programming: Python, JavaScript, ASP.net
Web Development: React, FastAPI, Node.js, HTML, CSS, TypeScript
Database: MongoDB, JSON, Postgres, SQL
APIs & Security: REST API, JWT
AI & Machine Learning: Machine Learning, Deep Learning, NLP, Computer Vision, Prodigy
Generative AI: OpenAI GPT-4, LLAMA, Google Gemini, Langchain, RAG, LLMs
Tools: Git, Jupyter Notebook, VS Code, Cursor
Education
Postgraduate - Full Stack Software Development
Lambton College
Mississauga, ON
May 2024 - Present
Bachelor's - Computer Science and Information Technology
Nepathya College
Manigram, Nepal
May 2017 - April 2021
Projects
Budgetly
- • Built a full-stack financial management platform with React/TypeScript and FastAPI microservices, enabling real-time analytics and Google OAuth integration.
- • Automated receipt extraction via GPT-4 Vision with 95%+ accuracy and fraud detection, cutting manual data entry by 80%.
IGBot
- • Automated Instagram profile and ID scraping using specific keywords, integrating OpenAI generated personalized messages for marketing outreach, resulting in a 25% increase in engagement and sending over 500 targeted DMs per week.
- • Deployed AI-generated responses to streamline Instagram DM marketing, leading to a 25% increase in engagement and boosting product visibility by sending 500+ targeted messages per week.
Digital Advertisement Classification
- • Developed a web application to categorize digital advertising data from multiple sources, facilitating advertisers to analyze campaign effectiveness and optimize strategies, leading to a 20% improvement in proposing accuracy.
- • Accomplished an accuracy rate of more than 90% in classifying advertisements by leveraging the power of TF-IDF for feature representation.
TaskFlow
- • Built a task management platform with automated reminders and real-time updates, resulting in 40% improved team productivity and efficient tracking of 100+ tasks weekly.
- • Developed collaborative workspace with task assignment and notifications, leading to enhanced team coordination and 30% reduction in missed project deadlines.
Awards
Top 5 Finalist - Startup and Idea Fest 2022
Awarded as the "Top 5 Finalist of Startup and Idea Fest 2022 Lumbini Province" by Honorable Minister Tilak Sharma.