Built a pricing optimization framework for a cafe menu using price elasticity of demand to find profit-maximizing price points while preserving demand.

• Problem & Impact: Focused on highly price-sensitive products (burgers, coke, lemonade, coffee), balancing margin and volume to avoid revenue loss and improve profitability across events, weekends, and weather conditions.
• What I Built: Loaded and joined product, transaction and calendar data (holidays, weekends, seasons, temperature); performed EDA and data cleaning; modeled price elasticity with regression (statsmodels OLS); derived demand curves and simulated revenue/profit across price points to identify optimal prices and actionable recommendations.
• Tech Stack: Python, Pandas, NumPy, Statsmodels, Matplotlib, Seaborn, Jupyter Notebook.

Built an end-to-end machine learning data pipeline to tackle truck delay prediction for a logistics use case, enabling data-driven decisions around on-time deliveries and operational efficiency.

• Problem & Impact: Designed the solution around delayed truck shipments, focusing on reducing operational costs, improving customer satisfaction, and optimizing route planning under varying traffic and weather conditions.
• What I Built: Set up relational data infrastructure on AWS RDS (PostgreSQL & MySQL); performed SQL-based analysis and data validation; configured an AWS SageMaker notebook environment; implemented EDA, preprocessing, and feature engineering; integrated Hopsworks Feature Store to create and serve reusable feature groups.
• Tech Stack: Python, SQL, NumPy, Pandas, Matplotlib, Seaborn, PyMySQL, Psycopg2, PostgreSQL, MySQL, AWS RDS, AWS SageMaker, Hopsworks, MySQL Workbench, pgAdmin4.

Creator • RGU, Aberdeen

• Problem & Impact: Built to help hotels, resorts and short-stay hosts deliver personalized, always-on guest support, automate routine workflows, and increase upsell revenue — all while keeping data on-prem for privacy and low latency.
• What I Built: Architected a modular LLM-agent platform with a SupervisorAgent that routes requests to specialized agents (CheckinAgent, RoomServiceAgent, MaintenanceAgent, EmergencyAgent, WellnessAgent). Fine-tuned a quantized Llama 3B model on synthetic guest dialogues for on-site deployment; integrated LangChain RAG pipelines for retrieval-augmented responses; implemented a realtime memory component to capture guest preferences and context-aware behaviour; and designed an ethical data-collection & retraining loop to continuously improve agents from anonymised usage data.
• Tech Stack: Llama 3.x (3B, quantized), LangChain, custom LLM agents, on-prem deployment, synthetic data generation, CRM integrations.

AI ML Teaching Assistant

• Supported instructors in delivering AI/ML course materials and assisted students with complex concepts.
• Provided guidance on practical implementation of AI/ML algorithms, tools, and techniques.

Product Intern

• Developed a Drowsiness Detection edge device (Raspberry Pi) using computer vision to alert distracted drivers.
• Created an OCR content extractor for structured & unstructured documents.
• Conducted load testing on servers using JMeter and Python scripts.