Developed AI-based solutions to detect police misconduct and bias in protest coverage using Computer Vision and Natural Language Processing (NLP). Built and annotated a custom video dataset for detecting police presence, violence, and officer identification using video classification and object detection models. Applied deep learning techniques (CNNs, Transformers) to classify protest footage and extract uniform details (e.g., collar numbers) from video frames. Conducted sentiment analysis on news and social media data to evaluate reporting bias in protest coverage. Delivered a fully functional prototype algorithm, tested on real-world footage and supported by statistical evaluation.

Conducted a two-month research project analyzing the competitive dynamics between Google and ChatGPT in AI development. Presented findings in a research paper session (Avlokan 2023). Strengthened research, analytical writing, and presentation skills.

Implemented a Markov Decision Process (MDP) solver to optimize Pacman’s gameplay strategy using reinforcement learning concepts. Designed algorithms for state evaluation, transition probabilities, and reward maximization. Conducted experiments to compare different policies and evaluated performance improvements in complex maze environments.

Analyzed the effect of redesigned store layouts on sales using real-world retail data. Applied hypothesis testing, regression modeling, and ANOVA to measure sales impact and customer behavior patterns. Interpreted statistical outputs to provide actionable business insights and visualized findings through Python-based data visualization techniques.

Developed efficient algorithms and applied object-oriented programming principles to solve real-world computational problems. Implemented data structures, recursion, and sorting algorithms for optimized performance. Practiced data manipulation using Python libraries (NumPy, Pandas) and evaluated algorithmic complexity through hands-on coding assignments.

Developed a data pipeline integrating COVID-19 datasets with natural language processing for text analysis and API integration. Implemented data cleaning, preprocessing, and transformation workflows. Designed interactive graphs and dashboards to visualize pandemic trends, enabling insights on case progression and public discourse.

Built a full data-mining pipeline, including data acquisition, preprocessing, feature engineering, and model training. Implemented clustering, classification, and association rule mining algorithms. Evaluated model performance using statistical metrics and presented results through visual analytics for decision-making support.

Implemented a Convolutional Neural Network (CNN) to classify handwritten digits from the MNIST dataset. Applied deep learning techniques such as convolutional layers, pooling, and dropout for improved generalization. Achieved high classification accuracy and analyzed misclassified instances to refine model robustness.

Designed and implemented a Support Vector Machine (SVM) classifier from scratch without relying on ML libraries. Extended the model to handle multi-class classification using one-vs-one and one-vs-rest approaches. Optimized hyperparameters and validated accuracy against benchmark datasets.

Developed an interactive data visualization dashboard using D3.js to represent complex datasets dynamically. Incorporated filtering, zooming, and tooltips for improved user engagement. Focused on usability and storytelling through data, enabling users to uncover hidden trends via visual exploration.