Professional profile
MSc Artificial Intelligence postgraduate at the University of Bath with hands-on project experience in Python, SQL, analytical pipelines, predictive modelling and applied AI. Built an interactive retail dashboard covering 9,993 orders, and evaluated fraud-detection models on 283,726 transactions. Previously supported Python programming labs as a Graduate Teaching Assistant, explaining technical concepts and debugging code.
Technical expertise
Programming & dataPython, pandas, NumPy, SQL, MySQL, Java, relational modelling, ETL, data validationAnalyticsExploratory analysis, KPI reporting, customer segmentation, time-series forecasting, Streamlit, Plotly; working knowledge of Power BI and TableauMachine learningscikit-learn, XGBoost, TensorFlow, Keras, class-imbalance handling, feature engineering, model evaluationAI & developmentFastAPI, RAG, MCP, Git, Jupyter, model evaluation and human-in-the-loop workflows
Professional experience
Graduate Teaching Assistant · University of Bath Nov 2025 – May 2026
- Supported Python programming lab sessions covering debugging, code review and data manipulation.
- Helped students investigate errors through systematic root-cause analysis and strengthen independent problem-solving skills.
- Explained technical subjects clearly to learners with different levels of programming experience.
Selected projects
E-commerce Sales Analytics & Interactive Dashboard
Python · pandas · Streamlit · Plotly · statsmodels
- Created reusable data-cleaning, KPI, forecasting and dashboard modules for 9,993 retail order lines.
- Identified a loss-making sub-category with $17.7K in negative profit and discount-driven margin risk in the sample data.
- Built a 12-month Holt-Winters forecast with 17.6% holdout MAPE; segmented customers using RFM.
Financial Fraud Detection Analytics
Python · scikit-learn · XGBoost · Streamlit
- Compared four classification approaches on 283,726 card transactions with a 0.17% fraud rate.
- Selected XGBoost (PR-AUC 0.82); tuned an evaluation threshold achieving 84% fraud recall.
- Reduced estimated cost in the project’s illustrative model by approximately 20% against the default threshold.
Cybersecurity SOC Copilot · Working MVP
Python · FastAPI · Streamlit · RAG · MCP
- Developed an offline-capable investigation assistant using synthetic alerts, auditable evidence and cited playbooks.
- Designed typed agent workflows, read-only MCP tools, approval-gated response proposals and PDF incident reports.
- Prototype only; no production deployment or production performance claims.
Cybersecurity Incident Analytics
Python · Streamlit · Data quality · KPI design
Analysed 18,000 synthetic incident records to explore security trends, incident severity and response performance.
Network Intrusion Detection · MSc Dissertation
Investigated machine-learning and deep-learning approaches, class imbalance, feature engineering and per-attack model evaluation.
Education
MSc Artificial Intelligence · University of Bath 2025–2026
Coursework in machine learning, deep learning, natural language processing and reinforcement learning.
BTech Computer Science Engineering · Amity University, Noida 2017–2021
Certifications
Oracle AI Foundation Associate · Oracle Generative AI Professional
Project metrics are based on analyses and controlled evaluations; they do not represent results achieved in a production business deployment.