Hi! I'm Sirjan

Context

Collections environments generate large volumes of call data but most operational teams have no structured way to monitor agent performance or call outcomes without manual digging. I built this project to demonstrate what that reporting infrastructure looks like when properly architected, using a synthetic dataset modeled on real industry behavior including time-of-day trends and holiday patterns.

Action

• Data Engineering: Used the Faker library in Python to generate a sophisticated dataset that mirrored industry trends.
• Business Logic: Programmed specific constraints for holiday based behavior and time of day trends to avoid statistical flatness.
• Visualization: Designed a Tableau dashboard with drill down capabilities to track RPC rates, payment conversions, and silence time.

Context

Research partner registrations were being managed entirely through email and spreadsheets. There was no way to track participation, match faculty systematically, or give leadership any view into what was happening across the program.

Action

• Centralized Hub: Built a SharePoint system to collect and organize all project proposals in a single, secure location.
• Smart Automation: Created Power Automate workflows to handle instant notifications and match faculty with external partners based on academic criteria.
• Executive Visibility: Designed Power BI dashboards to provide leadership with a real time view of project trends and engagement metrics.

Context

Generating accurate text descriptions for images requires a deep understanding of visual patterns. I developed this AI system to automate image captioning with a focus on model performance and security.

Action

• Model Fine Tuning: Optimized a BLIP architecture using a dataset of 20,000 images to improve description relevance.
• Performance Evaluation: Implemented a testing framework using BLEU scores to objectively measure the quality of generated text.
• Secure Deployment: Built an interactive Streamlit application featuring a session based authentication system for secure user access.

Context

Navigating graduate school applications involves significant uncertainty. I built this predictive tool to help applicants quantify their admission chances based on standardized academic metrics and research profiles.

Action

• Neural Network Architecture: Engineered a Keras regression model to process multi dimensional inputs including GRE, TOEFL, and CGPA scores.
• Streamlit Integration: Developed a web application that translates complex probability scores into intuitive categories like High Chance or Low Chance.
• Reproducible Research: Implemented a robust data pipeline and random seeding to ensure consistent, reliable predictions with an R squared score of 0.74.

Context

Real estate pricing is influenced by a complex mix of property features and market cycles. I built this tool to provide homeowners and investors with an objective, data driven estimate that considers both house specific details and the wider economic climate.

Action

• Ensemble Learning: Implemented a Random Forest regression model to handle the complex, non linear relationships between property tax, insurance, and sale prices.
• Feature Engineering: Incorporated economic indicators such as "Recession" and "Popularity" to refine predictions beyond standard physical dimensions like bedrooms and lot size.
• Interactive UI: Developed a Streamlit application with intuitive input forms and tooltips to ensure the model remains accessible and transparent for non technical users.