Bridging mathematical decision theory and fault-tolerant production systems.
I am Claudio Cabrera, a Sequential Decision Scientist and Machine Learning Systems Engineer specializing in autonomous e-commerce supply chains, multi-echelon inventory allocation, and algorithmic pricing.
Across the industry, a costly divide separates theoretical Operations Research models—which frequently remain trapped in brittle notebooks with lookahead bias—from production software engineering, which defaults to myopic SQL rules when making real-time pricing and routing decisions. My work eliminates that divide.
Grounded in quantitative and analytical practitioner experience across the University of Chicago Booth School of Business, EY, and BBVA, I architect end-to-end decision engines that formulate uncertainty mathematically via Warren Powell’s Universal Modeling Framework and execute in production behind strict Hexagonal boundaries, deterministic replay ledgers, and GAAP ASC 606 financial objectives.
Practitioner Pedigree
Institutional Foundation
University of Chicago Booth School of Business
Quantitative research and econometric modeling practitioner experience; rigorous empirical calibration, structural estimation, and high-dimensional statistical inference.
EY (Ernst & Young)
Enterprise analytical advisory, financial accounting alignment (GAAP ASC 606 / Activity-Based Costing), and risk-controlled operational modeling.
BBVA
Quantitative risk, stochastic modeling, and institutional data engineering under strict auditability and regulatory governance constraints.
Academic & Language Profile
Economics & Quantitative Methods coursework (Lima, Perú) · Bilingual Fluency in English (C2) & Spanish (Native).
Theoretical & Engineering Foundation
The Six Pillars
Every system in my portfolio is constructed at the intersection of six complementary disciplines—ensuring that mathematical optimality translates directly into resilient software and measurable P&L lift.
Powell’s Sequential Decision Analytics
Warren B. Powell (2022, 2026) · Universal Modeling Framework
Rigorous decomposition of every operational problem into State (Sₜ = Rₜ, Iₜ, Bₜ), Decision (xₜ ∈ Xₜ), Exogenous Information (Wₜ₊₁), Transition Physics (Sᴹ), and Objective Function (C), systematically benchmarking all four policy meta-classes (PFA, CFA, VFA, DLA).
Revenue Management & Capacity Control
Talluri & van Ryzin · Network Bid-Price & Allocation Theory
Multi-echelon inventory placement, Available-to-Promise (ATP) shadow pricing via Lagrangian duals, and capacity-constrained order routing across distributed fulfillment networks.
Contribution Margin & Pricing Optimization
Robert L. Phillips · Incremental Cost & GAAP ASC 606 Alignment
Dynamic B2B quote pricing and B2C markdown optimization grounded in exact unit economics—deducting Activity-Based Costing (ABC) fulfillment expenses, geodesic freight, and SLA stockout penalties.
Idiomatic Python 3.14 & Hexagonal Systems
Ports & Adapters · Alexis King’s 'Parse, Don’t Validate'
Zero-leakage domain isolation using frozen dataclasses with slots, structural pattern matching (match/case), strict MyPy typing, DuckDB local OLAP spines, and Hypothesis property-based invariant proofs.
Operational B2B & Marketplace CRM
Endogenous CLV, Churn State Evolution & Distributed Sagas
Treating Customer Lifetime Value (CLV) and churn propensity as endogenous belief-state variables that evolve with every fulfillment or rejection decision, backed by Transactional Outbox and Saga compensation guarantees.
Spatio-Temporal & Epistemic Econometrics
Wikle DSTM · Lohr Survey Sampling · Marzouk BOED · Chopin SMC
De-biasing geographic demand skew via Horvitz-Thompson estimators, probing stockout-censored demand with Active Learning, and tracking latent logistics delays via Sequential Monte Carlo particle filters.
Tooling & Infrastructure
Production Tech Stack
Sequential Decision & Math Core
Languages, Data & Analytical Spine
Distributed Architecture & QA
Cloud, Zero-Trust Security & MLOps
Currently Scaling: Project Aegis & Multi-Agent Supply Chain Orchestration
Calibrating a 4-policy tournament benchmark (PFA, BOED-tuned CFA, Dual Shadow-Price VFA, and ST-GCN Rollout DLA) over 100,000+ real Brazilian Olist e-commerce orders on a local DuckDB OLAP spine.
Ready to upgrade your decision architecture?
Available for high-impact B2B consulting sprints, fractional Principal Decision Scientist retainers, and global remote engineering roles.