Senior AI/ML Engineer

Building AI systems that generate$40M+ in business value

8+ years building production AI products for Fortune 500 companies — GenAI, recommender systems, and supply-chain optimization.

Explore My Work Let's Talk

In production

Multi-agent IVR

CVS Health · millions of patients

+10% CTR

Ranking at Walmart scale

10M+ users · p99 < 100ms

Trusted by teams at

$40M+
Business value delivered: 8+
Years of experience: Fortune 500
Enterprise clients: 20+
Production AI systems

Impact & Achievements

Transforming data into business value

Proven track record of delivering enterprise-scale AI solutions that drive measurable business outcomes

Driving ROI with AI

Delivered $40M+ in revenue through AI Market Mix Modeling, achieving 10% CTR improvement and 30% supply chain efficiency gains across Fortune 500 enterprises.

Revenue impact: $40M+
CTR improvement: 10%
Efficiency gains: 30%

From Pipelines to Production

End-to-end ML infrastructure architect specializing in scalable recommendation systems, GenAI solutions, and MLOps pipelines serving millions of users daily.

ML pipelines: 15+
Models deployed: 20+
Users served: 10M+

C-Level Strategy & Innovation

Bridging technical execution with strategic business outcomes, delivering data-driven insights that drive executive decision-making and digital transformation.

Strategic projects: 25+
C-suite presentations: 50+
Innovation labs: 5

My Journey

Experience & Education

Professional Experience

Jan 2026 – Current7 months

Senior Data Scientist & AI/Machine Learning Engineer

CVS Health

MinuteClinic IVR / millions of patients-45% response latency; -60% LLM token consumption

Key achievements

Architected and deployed a production-grade multi-agent IVR platform for CVS MinuteClinic using LangGraph StateGraph, enabling coordinated agent orchestration to resolve multiple patient intents within a single conversation.
Built and launched LocatorAgent using Mistral and LangGraph, designing deterministic middleware that reduced response latency by 45% and LLM token consumption by 60%, enabling millions of patients to discover locations.
Developed LiveTransferAgent, a real-time LLM routing and decision engine that classifies transfer intent and automatically deflects self-serviceable requests (appointments, billing, medical records, lab results).

Technologies

PythonLangGraphMistralMulti-Agent OrchestrationLLMs

Jan 2025 – Jan 20261 year, 1 month

Senior Data Scientist & AI/Machine Learning Engineer

Enero Group (OBMedia)

High-volume ads pipeline+5% RPC; +8% advertiser ROI via GenAI personalization

Key achievements

Boosted Revenue per Click (RPC) by 5% by modernizing keyword matching with a production-grade RAG system, architecting a Bi-Encoder retrieval layer indexed via FAISS to capture semantic user intent at millisecond latency.
Engineered a Multimodal Recommendation System utilizing Computer Vision (CLIP) to predict high-converting image-keyword pairings, automating creative selection and boosting CTR.
Drove an 8% uplift in Advertiser ROI by engineering a GenAI personalization engine, fine-tuning Transformer models to rewrite and optimize AdSense content relevance and increase user engagement signals.

Technologies

PythonRAGFAISSCLIPTransformers

Jan 2024 – Jan 20241 month

Data Scientist & Machine Learning Engineer

Walmart

Large-scale e-commerce platform+10% CTR; +25% Recall@K for cross-sell recommendations

Key achievements

Achieved a 10% increase in Click-Through Rate (CTR) by engineering a Learning-to-Rank objective within XGBoost, optimizing product ranking across billions of historical transactions rather than simple classification.
Increased Cross-Selling Recall@K by 25% by deploying an ML ranking architecture that fuses user history and item features to solve the cold-start problem in recommendations.
Enhanced recommendation coverage by developing BERT-based dense embeddings for product descriptions, allowing the pipeline to identify item similarities even for products with sparse interaction data.

Technologies

PythonSparkSQLXGBoostBERT

Jan 2023 – Jan 20241 year, 1 month

Data Scientist & Machine Learning Engineer

DHL Supply Chain

North America logistics operations+5% fleet utilization; +30% operational efficiency

Key achievements

Optimized fleet utilization (Backhaul Discovery) by 5% by designing a hybrid framework merging LightGBM with lag features for demand forecasting and linear programming for route optimization.
Improved operational efficiency by 30% across North American logistics networks by deploying predictive analytics for resource planning, directly reducing idle time and optimizing labor allocation.
Improved On-Time Performance (OTP) by engineering a production-grade ETA prediction model using XGBoost.

Technologies

PythonLightGBMXGBoostLinear ProgrammingSQL

Jul 2020 – Jan 20232 years, 7 months

Data Scientist

Huawei

Telecom commercial & network planning$40M revenue impact; +39% targeting efficiency

Key achievements

Drove $40M in incremental revenue and improved NPS by engineering a Geospatial Network Allocation model, identifying optimal locations for site expansion that maximized both commercial ROI and subscriber experience.
Outperformed legacy campaigns by 14% (boosting 5G targeting efficiency by 39%) by engineering an Uplift Modeling strategy (Causal Inference) that targeted customers with the highest incremental probability of conversion.
Reduced churn by 5% by building a Geospatial Risk Engine that segmented customer risk based on network quality.

Technologies

PythonSQLScikit-learnGeoPandasCausal Inference

Jul 2018 – Jul 20202 years, 1 month

Data Scientist

Conacyt

Municipal operations & routing30% route distance reduction via ML + optimization

Key achievements

Achieved a 30% reduction in waste collection route distance by combining XGBoost-based ETA prediction with routing heuristics and linear programming.

Technologies

PythonXGBoostGoogle OR-ToolsLinear ProgrammingGeospatial Analysis

Mar 2016 – Jul 20182 years, 5 months

Data Analyst

Baxter International Inc. (Fortune 500)

Manufacturing operationsOperational visibility via IoT manufacturing dashboards

Key achievements

Developed intuitive manufacturing dashboards in SQL to integrate IoT data (machines, sensors, and labor).

Technologies

SQLPower BIC#PythonIoT Data

Education

MSc in Optimization & Applied Computer Science

Focus on optimization and applied ML for real-world routing and planning problems.

BS in Computer Engineering

Strong foundation in algorithms, software engineering, and data structures.

Technical Expertise

Core Competencies

Deep expertise across the full ML lifecycle, from data engineering to production deployment

GenAI & LLM Systems

Multi-agent platforms, RAG pipelines, and fine-tuned models shipped to production.

LangChainLangGraphRAGVector DBsFine-tuningPrompt Engineering

Machine Learning

Ranking, recommendation, and forecasting models that move business metrics.

Scikit-learnXGBoostLightGBMPyTorchTensorFlowHuggingFace

Data Engineering & Big Data

Billion-row pipelines engineered for reliability, quality, and reuse.

PySparkDatabricksBigQueryAirflowdbtKafkaSQL

Cloud & MLOps

From notebook to monitored, low-latency production service.

GCPVertex AIAWSDockerKubernetesMLflowCI/CD

Analytics & Experimentation

Decision science and measurement frameworks executives can act on.

A/B TestingMarket Mix ModelingChurn & LTVCausal InferencePower BIPlotly

Optimization & Operations Research

Routing, planning, and resource allocation at supply-chain scale.

Google OR-ToolsLinear ProgrammingMetaheuristicsDemand Forecasting

Engineering Strategy

Engineering Philosophy

Principles guiding my approach to building production ML systems at scale

Production > Perfection

I prioritize shipping simple, explainable baselines (like XGBoost) over complex deep learning models to establish rapid feedback loops. A model in production beats a perfect model on a laptop.

Deploy simple baselines first (XGBoost/Linear) before complex models
p99 latency < 100ms and 99.9% uptime SLA as non-negotiables
Horizontal scaling with Docker for production resilience

Data-Centric AI

I focus 80% of effort on data quality and pipeline reliability (dbt/Airflow) rather than hyperparameter tuning. Clean data beats clever algorithms.

Versioned datasets with Git integration
Automated data quality checks (Great Expectations/dbt tests)
Schema evolution with backward compatibility

Business Alignment

I translate 'Mean Squared Error' into 'Revenue Lift' to align engineering efforts with company OKRs.

A/B testing framework for measurable impact
Revenue attribution models for ML features
Executive dashboards with business KPIs

The best code is the code that ships. The best model is the one that drives measurable business value.

Diego Hurtado

Case Studies

Featured ML Systems & Case Studies

Production ML systems delivering measurable business outcomes at enterprise scale

Gemma 3 1B Reasoning Fine-Tuning with Tunix (LoRA SFT -> GRPO)

Independent R&D (Google Tunix Hackathon / Kaggle TPU) · Jan 2026 - Present

Gemma 3TunixJAXLoRA SFTGRPO

Training had to improve reasoning depth while never violating a strict XML output contract under Kaggle TPU memory and cache constraints. Key failure modes included malformed tags, empty outputs, rollout concat mismatches, and KV-cache sizing errors during generation-heavy GRPO.

Business impact: 100% strict XML post-repair (40% raw -> 100% repaired)
Scale: Kaggle TPU-stable Tunix 0.1.6 pipeline

Read case study

Offline Multimodal Clinical Triage Copilot (MedGemma 1.5)

Independent R&D (Kaggle MedGemma Impact Challenge) · 2026 (Research / Demo Prototype)

MedGemma 1.5Multimodal AIComputer VisionFHIRDICOM

Clinical triage in low-connectivity and sovereignty-constrained settings requires image-grounded reasoning, prior-comparison evidence, and auditability without cloud dependencies. The workflow must prevent patient-context mismatches while still producing fast, clinician-readable outputs.

Business impact: ~10-14 min saved per complex case
Scale: Single T4 GPU, offline edge/on-prem

Read case study

Ranking + Recommendations at Scale (Walmart)

Walmart Global Tech · Jan 2024 – Dec 2024

PySparkGCPBERTRankingRecommendation Systems

Large catalogs require relevance ranking and cross-sell recommendations that generalize across products and changing user behavior.

Business impact: +10% CTR; +25% Recall@K
Scale: Billions of transactions

Read case study

Predictive Logistics + Optimization (DHL Supply Chain)

DHL Supply Chain · Feb 2023 – Jan 2024

PythonSQLXGBoostProphetLinear Programming

Logistics planning required better predictive signals and optimization to reduce inefficiencies and unlock backhaul opportunities.

Business impact: +30% efficiency; +5% backhaul
Scale: USA & Mexico operations

Read case study

AI Market Mix + 5G Targeting (Huawei)

Huawei · Sep 2020 – Feb 2023

PythonXGBoostGeospatialMarket Mix ModelingOptimization

Telecom commercial and network planning required data-driven models for expansion, pricing, and customer targeting.

Business impact: $40M; +39% targeting
Scale: Multi-market deployment

Read case study

Research Papers

Applied AI & Optimization Research

Peer-reviewed work bridging optimization and applied ML systems

Lecture Notes in Intelligent Transportation and Infrastructure, Springer · 2021

Waste Collection of Touristics Services Sector Residues Vehicle Routing Problem with Time Windows to an Industrial Polygon in a Smart City

Diego Hurtado-Olivares, José Alberto Hernández-Aguilar, Alberto Ochoa-Zezzatti, José Crispín Zavala-Díaz, Guillermo Santamaría-Bonfil

Presents an optimization framework for solid waste collection in smart cities, utilizing a heuristic algorithm that combines greedy initialization with simulated annealing for route improvement.

Smart CitiesVRPTWWaste ManagementMetaheuristics

Read paper

Lecture Notes in Intelligent Transportation and Infrastructure, Springer · 2021

Humanitarian Logistics for the Optimal and Timely Evacuation in High Buildings Within a Smart City Using an Adaptive Metaheuristic Context

Peter Savier Oropeza-Martínez, José Alberto Hernández-Aguilar, Alberto Ochoa-Zezzatti, Diego Hurtado-Olivares

Evaluates adaptive metaheuristic models for evacuating high-rise buildings during emergencies, incorporating heterogeneous crowd simulation to study dynamic agent behavior.

Humanitarian LogisticsEvacuationCrowd SimulationMetaheuristics

Read paper

Writing

Technical Writing

Articles on applied ML, modeling foundations, and practical data science systems

Mar 2026 · 13 min read

Google ADK vs LangChain vs LangGraph vs LangFlow vs LangSmith

A comprehensive breakdown of five AI agent framework tools, explaining how they serve different layers of the modern AI stack rather than competing directly with one another.

Read on Medium

Aug 2024 · 12 min read

Predicting Taxi Fare in New York City

Feature engineering, geospatial signals, and gradient boosting to build a robust pricing model.

Read on Medium

Mar 2023 · 17 min read

Customer Churn Prediction Using Machine Learning

A practical walkthrough of churn modeling and evaluation, including feature engineering and model selection.

Read on Medium

Testimonials

What People Say

Feedback from colleagues and leaders I've worked with

I worked side by side with Diego for more than a year. We made amazing things for Huawei, like creating a new methodology (Lifestyle Zones) for 4G sites optimization and 5G deployment - this project generated more than U…

Andres Garcia Jaramillo

Senior Manager, Business Consulting · Huawei Technologies

I had the opportunity to work with Diego as part of our Data Science team. He consistently delivered valuable results by aligning technical work with business impact. Diego stood out as a great team player—always willing…

Act. Miriam Elizabeth López

Lead Data Scientist · Walmart

I served as his thesis director during both his Computer Engineering and Master's studies. He demonstrated exceptional skills in AI, possessing a strong ability to bridge theoretical concepts with real-world applications…

Carlos Alberto Hernández Aguilar

AI Researcher (SNI Level III) · National Autonomous University of Morelos

I worked with Diego on Business analytics projects and taken advantage of his excellent analytical and modelling skills. Diego has a great ability to assemble data from multiple sources and apply logic and mathematical t…

Mohammed Wargui, Ph.D., MBA

Principal Business Consultant · Huawei Technologies

Diego supported and provided data science knowledge to state-of-the-art data-based projects for several Telcos in Latam. With his dedicated and diligent support, we delivered on-time innovative business solutions and met…

Paola Rozada

Managing Business Consultant · Huawei Technologies

Diego always performed in a mature professional way: strong technical skills in the creation of analytics algorithms and dashboards; enthusiasm, devotion and energy to execute his projects; good capability for insights d…

Felipe González Carrasco

Strategy Team · Huawei Technologies

I knew Diego Hurtado during our time together in Huawei - I can confirm his excellent human quality and professionalism when Data Science was involved into the discussions. I can say that he is a very good professional w…

Raúl Maldonado

Senior Business Architect · Huawei Technologies

I have the pleasure to count Diego as a colleague. He is a star in the rising. His comprehension on everything related to Data Analytics is far beyond his age. Invaluable contributions. Pristine work ethics. A true team…

Jose Miguel Fletcher

Operating Partner and Chairman · Huawei Technologies

Contact

Let's build production AI together

Open to senior AI/ML engineering roles and selected consulting engagements.

Email me Connect on LinkedIn