OQuant Data Science Consulting
OQuant Data Science Consulting logo

Turning data into defensible decisions

OQuant is an independent data science consultancy founded in March 2025 by Rodrigo Hermont Ozon — economist, econometrician, and PhD researcher in Systems and Industrial Engineering. The premise is simple: business decisions deserve the same rigor as peer-reviewed research. OQuant combines econometric modeling, statistical validation, and modern AI engineering to deliver analyses you can defend in front of a board, an auditor, or a regulator — not black boxes, and not slideware.

Services

What OQuant does

Forecasting & Time Series

Demand, price, and volatility forecasting with classical econometrics (ARIMA, GARCH-family, GAMLSS, Bayesian MSGARCH) and machine learning hybrids — with honest out-of-sample validation, not curve-fitting.

Econometric Modeling & Causal Inference

Answering "what actually drives this?" — regression diagnostics, panel and instrumental-variable methods, quasi-experimental designs, and Bayesian multicausal models grounded in economic theory.

Risk Analytics & Simulation

Quantitative risk measurement and scenario analysis: Monte Carlo simulation, extreme value modeling, portfolio and multi-objective optimization for pricing, sourcing, and investment decisions.

Machine Learning & Predictive Modeling

Supervised and deep learning models (including CNN+LSTM architectures and survival analysis for predictive maintenance) built with proper feature engineering, ablation testing, and MLOps discipline.

Business Intelligence & Decision Support

KPI frameworks, dashboards, and analytical pipelines that turn scattered data into a single, trusted view — from data modeling and governance documentation to executive-ready reporting.

Generative AI for Business

Practical LLM applications: internal chatbots, multi-agent and sub-agent workflow orchestration, and AI-assisted knowledge management — designed around your processes, with measurable value.

How to start

Engagement packages

Fixed scope, fixed timeline and a fixed fee agreed before any work starts. The fee depends on your data and your context, so every proposal is quoted individually.

Data & AI Diagnostic

The best first step: find out where data and AI can actually move your numbers.

  • Interviews with up to 4 people from leadership and operations
  • Inventory of the data you already have, and how far it can be trusted
  • 3 prioritized opportunities, each with estimated gain and effort
  • A 90-day plan and a 1-hour presentation to leadership

Timeline: 2 weeks

AI & Data Workshop for Managers

For leaders who need to decide where AI fits in the company, and where it does not.

  • Up to 25 participants
  • Hands-on exercise with your own data, anonymized
  • Material that stays with your team

Duration: 8 hours, over two mornings

Custom Forecasting

Demand, price, claims or default forecasts your planning can rely on.

  • One target variable and up to 3 data sources
  • Models compared against a simple benchmark, with out-of-sample testing
  • Uncertainty bands, not just a single number
  • Executive report, documented code and a simple dashboard
  • 2-hour handover training and 30 days of support

Timeline: 6 weeks

Reliable AI Assistant (pilot)

An assistant that answers questions about your documents or data, and never invents a number.

  • One use case and one data source (documents or tables)
  • 20 test questions you approve as the acceptance criterion
  • Every figure checked against the source before it is shown
  • Runs in your environment, under your control

Same method as the multi-agent commodity analyst.

Timeline: 6 weeks

Fractional Head of Data

Senior data leadership one day a week, without a full-time hire.

  • Priorities and roadmap for data and AI
  • Monitoring and recalibration of models in production
  • Mentoring for your analysts
  • Support on new questions as they come up

Monthly, minimum of 3 months

Request a quote for your project

Selected clients

Organizations OQuant has served

Grupo Fleury

Consulting work for one of Brazil's leading healthcare and diagnostic medicine groups.

Aurora Alimentos

Consulting work for one of Brazil's largest food industry cooperatives.

Cogna Educacional

Consulting work for one of Brazil's largest education companies.

Client engagements are confidential; sector references are shared with permission and project details remain under NDA.

How I work

A disciplined engagement model

Step 1

Diagnostic

Understand the decision, not just the data

We start with the business question and the decision it must inform. I audit the available data, map the constraints, and define success criteria before a single model is fit. If the data cannot support the question, you hear that on week one — not at delivery.

Step 2

Modeling & validation

Rigor before results

Candidate models are built, stress-tested, and compared against honest baselines with out-of-sample and backtesting protocols. Assumptions are documented, uncertainty is quantified, and every result comes with its limitations stated plainly.

Step 3

Deployment & handoff

Production code, not a PDF

Deliverables are reproducible: versioned code, documented pipelines, and dashboards or APIs your team can own. I train your people so the solution outlives the engagement.

Step 4

Monitoring & advisory

Models drift; decisions evolve

Optional ongoing support: performance monitoring, model recalibration, and advisory sessions as new questions emerge — so the analytics keep pace with the business.

Why OQuant

What you are actually buying

PhD-level rigor

Active doctoral research in multi-objective optimization, Bayesian econometrics, and time series at PUCPR, an MSc in Economic Development (UFPR), peer-reviewed publications, and reviewer roles for journals such as Applied Soft Computing. The methods behind your project are the ones surviving academic scrutiny.

18+ years of industry track record

Hands-on delivery inside large organizations — Volvo Group, BRF, CNH Industrial, Unimed — plus UNDP/UN research work and years of economic analysis for industry federations. I know how corporate data, politics, and deadlines actually behave.

End-to-end capability

From statistical theory to production code: econometric specification, Python/R/SQL implementation, cloud data pipelines (Azure, Databricks, PySpark), and MLOps. One consultant covers the full path from hypothesis to deployed system.

Volvo Group BRF CNH Industrial Unimed do Estado do Paraná UNDP Brazil FIEP
Next step

Request a quote for your project

Tell me in a few lines what decision you need to make, what data you have and when you need it. You will receive a proposal with scope, timeline and a fixed fee. The first conversation costs nothing, and it usually clarifies whether quantitative modeling can move your decision.

Book a 30-min call Email WhatsApp LinkedIn

contato.oquant@gmail.com

WhatsApp ↑