New
🤖
Sannu - AI Chatbot SaaS for UK SMBs
Full-stack SaaS product targeting UK dental practices and law firms. Built with Next.js 14, Supabase (pgvector), Clerk auth, Stripe billing across four tiers, and Anthropic Claude for streaming AI responses with RAG.
Next.js 14
TypeScript
Supabase
pgvector
Stripe
Anthropic SDK
New
🏢
UK Company Risk Monitor
Agentic pipeline consuming Companies House API to monitor and assess UK employer risk signals. FastAPI, async SQLAlchemy and Alembic, with a custom Claude tool-calling loop written directly against the Anthropic SDK — no LangChain. Full CI passing (ruff, mypy, pytest), Dockerised, deployed on Render against Neon Postgres.
FastAPI
Anthropic SDK
SQLAlchemy
Docker
Render
CI/CD
📊
Risk Factor Classification (SEC 10-K)
Automated pipeline to extract and classify risk factors from U.S. SEC 10-K filings. Compared Linear SVC, Naive Bayes, Neural Topic Modelling and LLMs (GPT-3.5-turbo), with the best configuration reaching 0.81 ROC AUC. Basis of a distinction-graded MSc thesis.
Python
LLMs
NLP
Hugging Face
Topic Modelling
🛡️
FNOL Intelligence System
ML system to predict insurance claim outcomes and costs from First Notice of Loss data. Deployed with Streamlit dashboards, Docker, and AWS with a retraining mechanism.
Scikit-learn
Streamlit
Docker
AWS
Feature Engineering
👥
Customer Segmentation - RFM Analysis
Customer segmentation of banking clients using RFM (Recency, Frequency, Monetary) analysis. Applied K-Means clustering to derive actionable segments for targeted marketing strategy and customer profiling.
Python
K-Means
Pandas
Scikit-learn
Clustering
📈
Cryptocurrency Price Forecasting
Time-series predictive models for cryptocurrency price movements that outperformed baseline forecasts using advanced feature engineering on historical market and sentiment data.
Python
Time Series
Feature Engineering
Forecasting
🏥
Hospital Bed Demand Prediction
ML system forecasting hospital bed demand from historical utilisation data, enabling proactive NHS capacity planning and operational decision-making. Full pipeline from EDA to deployment-ready model.
Python
Regression
Scikit-learn
Healthcare
🔒
Network Intrusion Detection
ML-based system to detect network intrusions and anomalous traffic patterns. Applied classification models on network traffic data to flag potential security threats with high precision.
Python
Classification
Cybersecurity
Scikit-learn
⚙️
Shift Performance Optimisation
Data analysis and predictive modelling to evaluate workforce shift performance from operational data. Identified patterns in productivity and inefficiencies to generate performance predictions and insights.
Python
EDA
Pandas
Scikit-learn
💰
Sales Cohort Analysis
Customer cohort analysis on transactional sales data to track retention rates, revenue patterns, and lifecycle behaviour over time - informing business strategy and marketing budget allocation.
Python
Cohort Analysis
Pandas
Visualisation
⚽
Women's Super League Analysis
Data analytics project examining performance trends, match statistics, and player patterns across Women's Super League seasons using Tableau dashboards and Python-driven statistical visualisation.
Tableau
Python
Sports Analytics
Visualisation
🏟️
Fan Attendance & Match Performance
Statistical study of the causal relationship between fan attendance and football match outcomes. Used hypothesis testing and regression to quantify the measurable effect of crowd presence on results.
Statistics
Hypothesis Testing
Regression
Python
✈️
Travel Assistant AI (Hotel Chatbot)
AI-powered conversational chatbot for hotel queries and booking assistance. Uses LangChain and LLM APIs for intent recognition, context management, and natural dialogue generation.
Python
LangChain
LLMs
Prompt Engineering