Data Science Consulting

Raw data.
Refined insight.

Refinaxa partners with organizations to transform complex data into clear, actionable intelligence. Specializing in machine learning, AI, and analytics that drive real decisions.

Where complexity
becomes clarity

Refinaxa is a boutique analytics and AI consulting practice built on a simple belief: the value of data is not in its volume, but in what you can learn from it.

This is a freelance practice focused on building real-world experience across industries — from finance and real estate to supply chain, sports, and education. Each project brings the same rigor and clarity regardless of domain, spanning unsupervised and supervised machine learning and applied AI, with a focus on models that are not just accurate but interpretable and actionable.

Every engagement is grounded in a real business problem. The goal is always the same: deliver analysis that decision-makers can understand and use, not just results that look impressive in a notebook.

What Refinaxa does

From exploratory analysis through production-ready models, Refinaxa helps organizations at every stage of their data journey.

01
Predictive Modeling

Classification, regression, and time series models built for interpretability and real-world deployment. Business questions are translated into modeling problems with results that make sense to stakeholders.

02
Segmentation & Clustering

Uncovering hidden structure in data through clustering, PCA, and association rule learning to segment customers, products, markets, and more into actionable groups.

03
Analytics & Reporting

Clear, audience-aware reporting that bridges technical analysis and business decision-making. Complex findings are translated into plain-language recommendations that drive action.

04
Neural Networks & Deep Learning

Autoencoders, LSTM forecasting, and neural network classifiers for problems where traditional methods fall short. Applied to demand forecasting, anomaly detection, and recommendation systems.

05
Data Pipeline & Preparation

Clean, analysis-ready data is the foundation of every good model. Preprocessing pipelines, feature engineering, and structured data preparation tailored to the problem at hand.

06
Agentic AI

AI systems that do not just answer questions but take action. Design and development of autonomous agents that reason across tools, APIs, and data sources to complete multi-step tasks with minimal human intervention.

Selected work

Applied machine learning across real business problems. Each project is grounded in a domain challenge and built end to end.

01
Real Estate
Melbourne Housing Price Benchmark

A portfolio project clustering 8,887 properties by physical characteristics to identify distinct property types for price benchmarking, independent of location labels.

K-Means Hierarchical t-SNE R
02
AI & NLP
Natural Language to SQL Query Engine

A portfolio project building an AI-powered query engine that translates plain English questions into executable SQL using GPT-3.5, applied to Amazon product and review data hosted in PostgreSQL.

GPT-3.5 PostgreSQL NLP Python
03
Investment
ETF & Mutual Fund Clustering

A portfolio project clustering Fidelity and Vanguard funds by actual behavior rather than category label to surface actionable options for long-term and retirement investors.

Clustering PCA Local LLM Python
04
Sports Analytics
WNBA Player Analytics & Betting Prediction

A portfolio project clustering WNBA players by playing style, predicting game outcomes, and identifying potentially mispriced betting lines using neural network models.

Neural Network Clustering Association Rules Python
05
Supply Chain
Inventory & Space Optimization

A portfolio project using LSTM neural networks to predict demand patterns, optimal reorder points, and warehouse slot assignments from historical transaction data.

LSTM Forecasting Clustering Python
06
Education
College Student Success Prediction

A portfolio project predicting at-risk students from early warning signals and recommending course sequences using association rules on historical enrollment patterns.

Association Rules Classification Clustering R

Technical toolkit

Python is the primary working language at Refinaxa, with additional experience in R and SQL. Every tool is chosen for the problem, not for novelty.

Unsupervised Learning
  • K-Means & hierarchical clustering
  • Principal components analysis
  • Association rule learning
  • Autoencoders
  • t-SNE & MDS visualization
  • Anomaly detection
Supervised Learning
  • Logistic & linear regression
  • Neural networks
  • LSTM time series
  • Classification modeling
  • Cross-validation & model selection
  • Feature engineering
Tools & Languages
  • Python (primary)
  • R
  • SQL & PostgreSQL
  • scikit-learn & TensorFlow
  • ggplot2 & matplotlib
  • Jupyter & VS Code

Start a conversation

Interested in working together or have a project in mind? Refinaxa is currently available for freelance engagements. Reach out through any of the channels below.

Send Message