Statistics · Machine learning · AI

We model what averages miss.

Most models are built for the average case. Real losses come from the tails: the stockout, the demand spike, the machine that fails, the loan that defaults. 4th Moment Analytics builds models that quantify uncertainty and hold up in production.

Start a conversation What we do

What we work on

Focused, senior, hands-on engagements, from problem framing to models running in production.

Forecasting & planning

Demand and volume forecasting with honest uncertainty ranges, built into planning cycles your teams actually run.

Risk & probabilistic modelling

Credit and default risk, time-to-event problems, simulation and scenario analysis for decisions made under uncertainty.

Customer & marketplace analytics

Pricing, churn, cross-sell, segmentation, recommendation and experimentation.

Industrial & IoT intelligence

Early failure warnings for equipment, anomaly detection and process quality prediction from sensor data.

GenAI & agentic systems

AI assistants and decision support grounded in your own data, with guardrails, evaluation and clear accountability.

Advisory & leadership

AI roadmaps, model governance, team set-up and fractional data science leadership.

How we work

Simple, measurable, built to last.

01

Frame the decision

Agree the business question, the success metric and how impact will be measured, before any modelling starts.

02

Build and prove

Start with strong baselines, add complexity only when it earns its place, and test on data the model has never seen.

03

Run in production

Reproducible code, monitoring and feedback loops, so models keep performing after go-live.

Selected experience

Work led or delivered by our founder across industries. Consulting engagements are described by sector to protect client confidentiality.

Pharma & consumer health

End-to-end supply chain planning

Demand planning and automation across countries and large product portfolios, running in monthly production cycles and feeding dispatch and logistics decisions.

Healthcare

AI-driven patient intake

An agentic AI system that guides patient intake and recommendations, with the decision logic kept deterministic and auditable.

Travel · Goibibo

Dynamic pricing at scale

Pricing, recommendation and image-quality models for a marketplace of 50,000+ hotels.

Public health · Aarogya Setu

National-scale risk modelling

Contributed to the risk modelling behind India's national contact-tracing app.

Banking · ANZ & Razorthink

Credit risk & customer analytics

Credit risk scorecards and customer analytics for banking across multiple geographies.

Manufacturing & operations

Industrial AI

Hospital inventory planning, early failure warnings for aviation ground equipment, process control in steelmaking and computer vision for infrastructure inspection.

11+ yearsin applied statistics, ML and AI
20+ member teambuilt and led
ISI KolkataB.Stat and M.Stat

Founder

Ankit Raj is a data science and AI leader trained in statistics at the Indian Statistical Institute, Kolkata. He most recently led data science at an AI consulting firm in Bengaluru, where he built and led a team of more than 20 data scientists and engineers across concurrent engagements in pharma, healthcare, consumer goods and manufacturing, while staying hands-on in solution design.

Earlier, he built dynamic pricing at Goibibo, contributed to risk modelling for the Aarogya Setu app, and developed credit risk and customer analytics in banking at ANZ and Razorthink.

Have a decision that depends on the edges of your data?

Tell us about the problem. We will tell you honestly whether data science can help, and how we would approach it.

Email us