Hi, I'm Balaaparajitha Senthilkumar
I am an Integrated M.Sc. Data Science student with a strong foundation in machine learning, deep learning, and applied artificial intelligence. I have hands-on experience developing efficient and scalable AI solutions across domains such as natural language processing, reinforcement learning, and time-series analysis.My academic journey includes international exposure through a student exchange program at the University of Barcelona, where I gained global perspectives on data-driven problem solving and collaborative research. I enjoy working at the intersection of theory and real-world application—optimizing models, experimenting with modern architectures, and transforming complex data into actionable insights that support informed decision-making.
SKILLS
What I work with
AI & Machine Learning
Machine Learning · Deep Learning · Model Optimization · Reinforcement LearningData & Analytics
Data Analysis · Time Series Forecasting · Data Visualization · Business AnalyticsNatural Language & Vision
Natural Language Processing · Computer VisionFoundations
Data Structures · Big Data Fundamentals · Statistical Thinking
How I build
Programming
Python · SQLML & Data Libraries
NumPy · Pandas · Scikit-learn · TensorFlowVisualization & Notebooks
Matplotlib · JupyterData Collection & Big Data
BeautifulSoup · Hadoop
Academic Journey Across Institutions
Education
2025 — International Exchange
Universitat de Barcelona, Spain
Explored global perspectives on data-driven problem solving through an international student exchange program.2022 – Present
Integrated M.Sc. in Data Science
Strong foundation in machine learning, deep learning, and applied AI.
CGPA: 8.11 / 102020 – 2022
Higher Secondary Education
Percentage: 85.8%2015 – 2020
Secondary Education
Percentage: 90%
Key Projects
PROJECT 1
Time-Efficient Fine-Tuning with LoRA
This project focuses on improving the efficiency of fine-tuning transformer models by implementing Low-Rank Adaptation (LoRA) on DistilBERT. Instead of updating all model parameters, LoRA enables parameter-efficient fine-tuning by training only a small subset of low-rank matrices, significantly reducing computational cost. The approach was applied to the SST-2 sentiment classification task, achieving near-baseline accuracy while substantially lowering training time and resource usage, demonstrating the practical benefits of efficient model adaptation.Technologies
Python · Transformers · DistilBERT · LoRA · NLP
PROJECT 2
EV Charging Dynamic Pricing Agent
In this project, a reinforcement learning-based dynamic pricing agent was developed to optimize electric vehicle charging costs and enhance user experience under varying demand conditions. The charging environment was modeled as a sequential decision-making problem, with a custom reward function balancing cost efficiency and user satisfaction. By incorporating simulated real-time inputs such as energy demand and weather conditions, the agent dynamically adjusted pricing, showcasing how reinforcement learning can enable adaptive and scalable solutions for smart energy systems.Technologies
Python · Reinforcement Learning · Simulation Models · Data AnalyticsOverview
PROJECT 3
Stock Market Price Prediction (NVIDIA)
This project explores long-term stock price prediction using deep learning by transforming time-series data into frequency-domain representations. Using 26 years of historical NVIDIA stock data, Short-Time Fourier Transform (STFT) was applied to generate spectrograms, which were then used to train a CNN-based forecasting model. The system successfully captured temporal patterns, generated buy/sell signals, and was evaluated using Mean Absolute Error (MAE), highlighting the effectiveness of combining signal processing with deep learning for financial time-series analysis.Technologies
Python · CNN · Time Series Analysis · STFT · Financial Data Modeling