You know how everyone talks about "data-driven decisions" but nobody actually knows where to start? So what I do is build tools that bridge that gap — from customer churn prediction models to AI-powered analytics. In fact, I've spent 10+ years in customer analytics and now build the AI tools I wish I had back then.

🛠️ Stack: Claude Code · Hermes · Codex (AI) | Python · R · SQL (Data) | Firebase · Vercel (Infra)

💼 Consulting: Founder at datafying, helping businesses with customer analytics (churn prediction, segmentation) and turning data into decisions that elevate customer experience

👨‍🏫 Mentoring: 10+ emerging data professionals to become strategic business partners who drive decisions with data

🔨 Currently: AI-powered customer analytics tools and agentic workflows

📫 Let's connect — X/Twitter · LinkedIn

• mimo-storyteller — Multi-character TTS audio story app with karaoke highlighting
• mimo-reader — Design custom voices, clone voices, and generate natural speech in your browser
• qwen3-tts-voice-clone — High-quality voice cloning powered by Qwen3-TTS — zero-shot from a single reference audio
• qwen3-asr — Audio transcription with Qwen3-ASR — automatic language detection, word-level timestamps, batch processing
• persona — Synthetic persona generation and interview skill for Claude Code
• melbourne-property-intelligence — LLM-powered property market intelligence with RAG, FastAPI, and Streamlit

• autokeras-text-classification — Spam detection and product review analysis with zero architecture decisions
• pycaret — Predicting employee attrition — comparing 10+ classifiers with SHAP interpretability
• lazypredict — Benchmarking every ML algorithm in one line of code
• prophet-time-series — Airline demand forecasting with seasonal decomposition and confidence intervals
• regression-using-tensorflow — Predicting fuel efficiency with a TensorFlow/Keras neural network
• review-data-using-SVC — Sentiment classification of restaurant reviews — full NLP pipeline
• dython — Measuring associations between categorical variables with Theil's U heatmaps
• google-cloud-vision — Image labelling, face detection, and visual search via Cloud Vision API

• ml-with-r — Predicting diabetes with GLM — full evaluation with accuracy, recall, precision, F1, AUC
• kaggle-predict-house-prices — Kaggle competition — XGBoost on 80 features with modelStudio explainability
• predict-marketing-response-with-xgboost — XGBoost for marketing campaign response — customer segmentation
• tidymodels — Decision Tree vs Random Forest vs XGBoost — progressive comparison with tuning
• model-studio — Comparing model explainability across algorithms with interactive DALEX dashboards
• linear-regression-in-r — Simple vs multiple regression — Facebook has 4x the per-dollar return of YouTube
• logistic-regression-in-r — Three datasets, coefficient interpretation, and probability math
• ggtext — Rich text in ggplot2 — colour-highlighted titles and styled axis labels

• puff — Progressive web app that helps you quit smoking by simulating the cigarette experience
• ai-for-kids — Fun, interactive course teaching kids aged 5-7 about Artificial Intelligence
• how-to-spell — Voice-powered spelling helper for kids — tap the mic, say a word, see it spelled
• sql-for-everyone — Interactive SQL course for non-technical business professionals
• claude-code-dashboard — Interactive token usage dashboard for Claude Code
• image-tinder — Swipe through photos like Tinder — keep the good ones, ditch the rest
• slaps — Two-player face-to-face card game for mobile browsers
• noise-monitor — Real-time noise level monitor with personalised spoken alerts
• property-auction-dataviz — Melbourne property auctions, mapped — scrape, clean, geocode, and explore on an interactive Leaflet map