Technical Projects
Sequence Modeling with LSTM for Stock Market Forecasting
This project involved analyzing 10 years of historical stock market data to build a predictive model for forecasting future stock closing prices. The model was developed from scratch using Long Short-Term Memory (LSTM) architecture implemented with standard Python libraries within the GWU Neural Network Library. By capturing sequential patterns in time series data, the model provided valuable insights into market trends and demonstrated strong performance in test case evaluations. click here
Exploring Domain Adaptation in Language Models Through Next Sentence Prediction
This project involved evaluating the performance of pre-trained language models on the Next Sentence Prediction (NSP) task across domain-specific datasets. By comparing ALBERT and GPT-2, the study assessed their ability to generalize from a general-purpose corpus to specialized domains such as healthcare, journalism, and e-commerce. The models were analyzed using accuracy and perplexity metrics, providing insights into their adaptability and effectiveness in understanding domain-specific language. click here
Exploring the Suitability of Cosmos DB for Scalable NoSQL Applications
This project involved evaluating Microsoft Azure Cosmos DB as a NoSQL data storage and analysis solution. The study explored its functionality in data exploration scenarios and assessed its strengths and limitations compared to other database technologies, offering insights into its suitability for scalable, cloud-based applications. click here
