Data Science Machine Learning

AAYUSH DONGOL

***************@*****.*** 203-***-**** LinkedIn Medium GitHub

EDUCATION

University of New Haven (UNH) – Tagliatela College of Engineering West Haven, CT Masters in Data Science Awarded Scholarship GPA: 4.0 / 4.00 Aug 2024 - May 2026 Coursework: Introduction to Programming (Python), Math for Data Science, Intro to Data Science, Machine Learning, Intro to Artificial Intelligence, Distributed & Scalable Data Engineering, Natural Language Processing, Deep Learning, Computer Vision, Power BI.

Prime College (BIM) Kathmandu, Nepal

Bachelor’s in Information Management Academic Scholarship GPA 3.47/4.00 Oct 2017 – Apr 2022 Coursework: Structured Programming, Digital Logic Design, Computer Information System, Data Communication and Networking, Java Programming, Web Programming, Database Management System, Artificial Intelligence (AI), Software Project Management, Data Mining and Warehousing, Object Oriented Analysis. SKILLS

Languages: Python, JavaScript, Typescript, SQL

Frameworks: Pandas, Numpy, Scikit-Learn, NLTK, PyTorch, Tensorflow, OpenCV, YOLO, Gensim

Web-Frameworks: ReactJS, NextJS, ReactNative, NodeJS

Database & Cloud: Firebase, MongoDB, MySQL, AWS

Design & Tools: Power BI, Excel, Photoshop, Canvas, Lightroom

Version Control: Git/GitHub, GitLab

WORK EXPERIENCE

Youth Innovation Labs, NGO Kathmandu, Nepal

Software Developer – Mid Level Nov 2022 – Aug 2024

Designed and implemented highly interactive front-end solutions using ReactJS, Redux, TypeScript, and Mapbox-GL, delivering real-time geospatial data visualization and enhancing decision-making capabilities.

Developed a multilingual video player with adaptive streaming, ensuring seamless user experiences across diverse language preferences and accessibility needs.

Developed responsive and mobile-friendly user interfaces, optimizing usability and engagement across multiple devices and screen sizes.

Led the performance optimization of mission-critical platforms like Bipad-Portal, NDRRMA, and Shikshya, enhancing data accuracy, system reliability, and response times.

Engineered intelligent state management solutions with Redux, reducing API calls and boosting application speed by 40%.

Introduced a customized geospatial analytics dashboard, allowing government agencies to track climate risks, disaster trends, and emergency response effectiveness in real time.

Collaborated closely with cross-functional teams, including designers, data scientists, and policymakers, to develop intuitive and impactful digital solutions tailored to real-world challenges.

Conducted code reviews and performance optimizations, maintaining best practices and ensuring scalable, maintainable, and secure applications.

Integrated real-time mapping features with Mapbox-GL, enabling location-based insights and interactive visualizations for improved disaster response and planning.

Led the implementation of progressive enhancements and accessibility standards, improving user engagement by 35% ensuring an inclusive and high-performance user experience for all audiences. ThemeGrills, (WordPress Based) Kathmandu, Nepal

Associate Software Developer Internship Dec 2021 – Nov 2022

Enhanced an LMS system by developing key-features, including a custom text editor and an interactive quiz section, improving user engagement and content creation.

Built dynamic front-end components using React and Typescript, delivering a responsive and feature-rich e-learning platform.

GrowByData Kathmandu, Nepal

Data Cloud Internship Nov 2019 – Dec 2021

Worked on Data Migration, Data Quality Assurance. ACADEMIC PROJECTS

Predictive Models

Implemented a Feedforward Neural Network in PyTorch for binary diabetes classification, utilizing 8 input features, ReLU activation hidden layers with optimizing functions.

Implemented advanced data preprocessing techniques, including data cleaning and normalization, and employed CrossEntropyLoss for reliable performance evaluation in predictive modeling with an accuracy of 83%. Periodic evaluations were made on a separate test set to monitor the models performance. GitHub Deep Learning Models (Computer Vision):

Developed and deployed a CNN model using TensorFlow for plant disease classification, achieving 93% training accuracy with a dataset of 100,000+ labeled images.

Leveraged data augmentation (rotation, flipping, scaling) and image preprocessing (rescaling, normalization) to enhance model robustness and prevent overfitting.

Integrated Flask API for seamless backend functionality, enabling real-time plant disease prediction.

Built a ReactJS frontend for a user-friendly interface, providing real-time results and predictions.

Implemented transfer learning with pre-trained models (VGG16, ResNet) to enhance performance and reduce training time.

Generative Models GitHub

Fine-tuned GPT-2 on a custom dataset, achieving a 35% reduction in perplexity, optimizing text coherence and fluency using a custom tokenization function by including special token (eos_token). Customized the pipeline and fine-tuning to align with specific requirements.

Custom Multi-Parameter Dataset for Plant Disease Diagnosis & Chat-Based AI

Engineered a multi-modal dataset combining image, environmental (temperature, humidity), and textual data for plant disease classification.

Collected and annotated 10,000+ labeled plant disease images with symptoms, severity levels, and recommended treatments.

Structured the multi-parameter dataset into a format suitable for training a chat-based AI model, mapping symptom- based textual inputs to relevant disease predictions and responses.

Developed a Transformer-based chatbot (using GPT-2/BERT) for real-time disease diagnosis, achieving 90%+ accuracy in response relevancy.

Contact this candidate