Hi, my name is
Sumant Kumar.
I'm a Data Science Enthusiast
with a keen interest in exploring, analyzing, and deploying data-driven solutions. I’m passionate about leveraging technologies like Computer Vision, Natural Language Processing, and Generative AI to solve real-world problems.
01.About Me
Hello! My name is Sumant, and I am passionate about leveraging the transformative power of Data Science across domains such as Computer Vision, Natural Language Processing, Generative AI, and other advanced AI technologies. I enjoy working with data exploring, visualizing, building, and deploying models to deliver impactful solutions.
I recently completed my Master of Computer Applications (MCA) from the Central University of Karnataka, graduating with an aggregate of 88%. Alongside my studies, I gained valuable experience through a six-month Data Science internship at IIT Jodhpur. During this internship, I worked on advanced topics such as aggression detection and multilingual datasets.
Additionally, I have earned certifications in Machine Learning Specialization and Deep Learning Specialization from DeepLearning.ai, further enhancing my expertise. I am now eager to apply my skills in a challenging environment where I can contribute to driving innovation and organizational growth.
Here are my skills:
- Languages: Python, C/C++ (Basics)
- Databases: Oracle Database, PostgreSQL, MySQL, MongoDB
- Data Science: Statistics, Machine Learning, Deep Learning, Digital Image Processing, Natural Language Processing
- Data Analysis Tools: Matplotlib, Seaborn, Plotly, Dash, Streamlit, Power BI, Tableau, Excel
- MLOps Tools: MLFlow, GitHub Actions, DVC, Docker, Airflow, AWS Cloud (Learning Phase)
- Libraries: NumPy, Pandas, Scikit-learn, OpenCV, NLTK, Transformers (Hugging Face)
- Developer Tools: Jupyter Notebook, Google Colab, Kaggle, Visual Studio Code, PyCharm, GitHub
- Deep Learning Frameworks: TensorFlow, Keras, PyTorch
- Web Technologies: HTML, CSS, JavaScript, Flask, Django (Learning Phase)
02.Where I’ve Worked
Data Science Intern at IIT Jodhpur (Social Network Analysis and
Application Lab)
Feb. 2024 – Jul. 2024
- I contributed to the "Indian Multi-lingual Aggression Dataset Development and Analysis" project. I worked on annotating and analyzing texts to identify aggressive behavior.
- I was involved in implementing advanced machine learning models for aggression detection, utilizing transformer-based models such as BERT and RoBERTa for classification tasks. I also gained hands-on experience with data preprocessing and fine-tuning pre-trained models to improve accuracy in identifying aggressive content.
- In addition to aggression detection, I contributed to the development of multi-lingual topic modeling using statistical methods and Large Language Models like BERTopic. This allowed us to extract meaningful topics from datasets, which enhanced the model’s understanding of diverse language patterns.
- Throughout this internship, I applied a wide range of skills, including data annotation, model evaluation, and large-scale data analysis. I worked extensively with tools such as Python, NLTK, transformers (Hugging Face), and TensorFlow, further refining my expertise in Deep Learning, machine learning, and NLP.
03.Some Things I’ve Built
Machine learning (ML)
A Novel Approach for Image(Fruit) Classification using Learnable Weighted Skip Connections in Depthwise Separable Convolution
Proposed a novel approach for image classification called Learnable Weighted Skip Connections, which increases classification accuracy and overcomes the vanishing gradient problem.
Nail Disease Detection using Image Processing and Machine Learning Techniques
The project involved the classification of eight distinct nail diseases by leveraging advanced image processing and machine learning techniques. Implemented Gabor, Gaussian and Variance filtering to accurately extract features from nail images. Achieved 85% accuracy with Random Forest.
Twitter Post Aggression Detection and Classification
Developed a classification system to detect and categorize Twitter posts(English) into two classes: Aggression (AG) and Non-Aggression (NAG). Implemented Naive Bayes as a baseline model, leveraging TF-IDF features and achieving an accuracy of 90%. Applied zero-shot classification using Hugging Face’s Transformers.
Get In Touch
I am actively seeking new opportunities and always open to connecting with like-minded professionals. Whether you have an exciting project, collaboration, or just want to chat about data science and AI, feel free to reach out. I’ll do my best to respond as quickly as possible!
Say Hello