Data & AI Solution Engineering
Framing needs, shaping requirements and designing integrated solutions from architecture through delivery.
Data & AI Solutions Engineer · Applied AI · Intelligent Automation · Systems Architecture
Evelyn Gutierrez, PhD
Complex problems.
Practical Data & AI solutions.
I turn complex and ambiguous problems into practical Data & AI solutions.
With more than a decade across data, machine learning and AI, I work from problem framing and architecture through hands-on implementation, validation and industrialization.
A CONNECTED TRAJECTORY
Statistics → Machine Learning → Computer Vision → Industrial AI → Intelligent Automation → Data & AI Systems
01 / Expertise
Deep technical expertise.
Choosing the right approach means understanding the problem, the evidence and the environment where a solution needs to work.
Applied AI & Intelligent Automation
Computer vision, multimodal AI, GenAI, retrieval and workflow automation selected according to the problem.
Data & AI Systems
Data pipelines, cloud, distributed processing, APIs, databases and observability for operational solutions.
Technical Problem Solving & Leadership
Technical decisions, trade-offs, mentoring, stakeholder alignment and the path toward industrialization.
02 / Selected work
Applied problems.
Evidence that matters.
Selected contributions across industrial AI, knowledge systems and research. Industrial projects are anonymized.
Data & AI Solutions Engineer
Systems · Architecture · Automation
From opportunity framing and solution design to evaluation, integration and production-oriented delivery.
01 / Industrial computer visionAI solution · Production validation
Making visual inspection
work on the production line.
Problem: Automate visual quality inspection in a production environment while handling rare defects, evolving labels and class-specific performance.
My role: Problem framing · Data strategy · Model development · Evaluation design · Production validation
>95%on relevant evaluation metrics
~1/minvehicle cadence in the production context
Several classes reached around 99%. Results are class- and metric-dependent; line cadence describes the operating context, not measured model latency.
Architecture, decisions & evidence
Approach: Defined an image-sampling and annotation strategy combining representative data with negative cases, then iterated on labels as model evidence exposed gaps.
Key decisions: Used class-specific thresholds, error analysis and semi-automatic out-of-time testing instead of relying on one aggregate score.
Result: Deployed models validated on real production cases, with several classes near 99% AP and targeted iteration for weaker classes.
02 / 3D & geospatial data systemsData systems · Cloud processing
From heterogeneous data
to digital-twin pipelines.
Problem: Turn heterogeneous 3D and geospatial sources into scalable, reproducible pipelines that can support operational digital-twin use cases.
My role: System design · Distributed data processing · Technical trade-offs · Industrialization
Multi-source3D, LiDAR, terrain, OSM and satellite data
Operationalmodular pipelines designed for scale and traceability
Architecture, decisions & evidence
Approach: Developed distributed processing with Databricks, PySpark and Delta Lake, supported by PostgreSQL/PostGIS data layers and containerized Azure workloads.
Key decisions: Evaluated accuracy, compute cost and processing time while introducing logging, metrics, traceability and reproducibility.
Result: Modular operational pipelines and a clearer transition from research methods to engineering delivery across product, simulation and cloud teams.
03 / Knowledge & agent systemsRAG · Agent automation
Designing knowledge workflows
around real decisions.
Problem: Support technical research, knowledge access and document-heavy decisions without treating an agent or LLM as the default solution.
My role: Use-case framing · Architecture · Agent orchestration · Evaluation · Technical direction
Modularreusable layers for interaction, orchestration, tools and data
Human-guidedfeedback and iterative review built into agent workflows
Technical choices were assessed against source quality, commercial-use constraints, context reliability and the needs of intended users.
Architecture, decisions & evidence
Approach: Compared scientific and web sources, indexing strategies, chunking approaches and lightweight multimodal options for retrieval.
Key decisions: Designed modular agent layers, tested persistent memory and human-in-the-loop behavior, and evaluated Dockerized local-model deployment.
Result: Reusable system components and clearer governance criteria for moving beyond isolated demonstrations.
04 / Applied AI discovery & automationOpportunity framing · Prototyping
Choosing the right automation
for the workflow.
Problem: Identify where automation could create practical value across technical training, document quality and engineering work.
My role: Workflow analysis · Solution selection · Prototyping · Expert validation
3 workflowsneeds, constraints and adoption barriers analyzed with domain teams
Fit for purposesemantic search, deterministic rules or AI agents selected by need
Architecture, decisions & evidence
Approach: Mapped workflows and built focused demonstrators with domain experts.
Key decisions: Chose semantic search for knowledge access, deterministic rules for PDF checks and AI-assisted prototyping where uncertainty justified it.
Result: A knowledge-assistant demonstrator, a tested PDF-validation prototype and a framed engineering proof of concept ready for expert-led next decisions.
05 / Medical imaging & applied researchAI R&D · Emerging solutions
Structuring emerging AI
into rigorous R&D.
Problem: Structure emerging medical-imaging and synthetic-data approaches into rigorous, decision-ready R&D.
My role: Research framing · Experimental design · Pipeline development · Technical mentoring
End to endfrom evidence and datasets to experimental pipelines
Decision-readyevaluation and maturity milestones made explicit
Architecture, decisions & evidence
Approach: Designed end-to-end experimental pipelines and explored conditional generative approaches combining diffusion models and Gamma-VAE concepts.
Key decisions: Defined evaluation and maturity milestones while using research assistants for evidence, market, funding and dataset discovery.
Result: Structured engineering work, explicit evidence gates and technical communication suitable for both specialist and non-specialist stakeholders.
06 / PhD · Multimodal computer visionResearch depth · Multimodal AI
Bringing geometry
and temperature together.
Problem: Combine color, temperature and geometry into a practical multimodal system for chronic-wound monitoring.
My role: Research design · Multimodal pipeline · Clinical coordination · Publication
RGB + thermalmultimodal information registered on 3D geometry
4,000+ imagesinternational clinical data collection coordinated
Architecture, decisions & evidence
Approach: Designed the acquisition-to-visualization pipeline across RGB, thermal imaging, 3D reconstruction and segmentation.
Key decisions: Prioritized portable devices, repeatable acquisition and clinically meaningful outputs.
Result: International clinical data collection involving more than 4,000 images and peer-reviewed publications.
03 / How I work
From the real need
to a usable system.
I contribute across the AI lifecycle, with evidence and domain feedback shaping each next step.
- 01
Understand
Clarify the users, constraints, available data and success criteria.
- 02
Structure
Turn ambiguity into a tractable problem and identify the highest-value use cases.
- 03
Design
Compare approaches and define components, interfaces and evaluation strategy.
- 04
Build
Develop pipelines, models, agents, APIs and demonstrators hands-on.
- 05
Validate
Measure performance and collect expert feedback against realistic needs.
- 06
Industrialize
Work with engineering teams on reproducibility, integration and operational constraints.
An iterative process: what we learn in validation informs what we build next.
04 / Experience & direction
A broader scope.
The same problem-solving thread.
Quantitative foundations, research depth and applied engineering—now converging in AI systems and technical direction.
2023–Present
Applied Scientist / Senior Data Scientist · Expleo Group
Data & AI technical contributor and lead across industrial, R&D and innovation work. My scope spans problem framing, data and system architecture, hands-on implementation, evaluation, industrialization and technical mentoring.
Current work combines data investigation, cloud and distributed systems, industrial computer vision, intelligent automation and applied research—with increasing responsibility for technical trade-offs, reusable components and the transition toward operational systems.
See selected contributions2019–2023Dual PhD · 2023
3D computer vision & multimodal imaging
Doctoral work across computer science and engineering: combining 3D models, thermal imaging and deep learning for chronic wound assessment, while coordinating clinical data acquisition and supervising students.
Université d’Orléans · Pontificia Universidad Católica del Perú
Earlier experience2011–2018
Statistics, machine learning & decision support
Credit risk modeling, geospatial analysis and consulting. Building a foundation in quantitative reasoning, varied data sources and applied business problems.
Earlier roles
- Data Scientist — Credit Risk Modeling Specialist
LenddoEFL · March 2015–October 2018 - Geo-Intelligence Consulting Analyst
Business Analytics SAC · January 2014–March 2015 - Credit Risk Analyst
Entrepreneurial Finance Lab · September 2011–December 2013
05 / Capabilities
Tools follow
the problem.
I choose the architecture and technical approach that fit the problem, evidence and operating constraints.
Applied AI
Computer vision · Multimodal AI · LLMs · RAG · Agents
PyTorch / TensorFlow / OpenCV / Open3D / CrewAI / LlamaIndex
Data & AI Systems
Data pipelines · APIs · Retrieval · Distributed processing · Automation · Databases
SQL / PostgreSQL / PostGIS / BigQuery / PySpark / Delta Lake / FastAPI
Platforms & Engineering
Cloud platforms · Containers · CI/CD · Geospatial tooling · Operational integration
Python / Azure / GCP / Databricks / Docker / GitLab CI/CD / GeoPandas
06 / Research & publications
Scientific depth.
Applied perspective.
My research background brings rigor to experimentation, model evaluation and the interpretation of results.
Education
Two disciplines.
A shared research question.
- PhD in Computer Science
- Université d’Orléans
- PhD in Engineering
- Pontificia Universidad Católica del Perú
- MSc in Statistics
- Pontificia Universidad Católica del Perú
- BSc in Statistical Engineering
- Universidad Nacional de Ingeniería, Peru
Computer vision · Deep learning
Multi-View Data Augmentation to Improve Wound Segmentation on 3D Surface Model by Deep Learning
Read publication Multimodal imagingMultimodal and Multiview Wound Monitoring with Mobile Devices
Read publication Thermal imaging · Clinical monitoringCorrection of Temperature Estimated from a Low-Cost Handheld Infrared Camera for Clinical Monitoring
Read publication
Evelyn GutierrezStatistics → Data → AI → Systems
07 / About
I enjoy problems that require
both analysis and building.
I design and build Data & AI solutions for complex technical and operational problems, working across data, models, software, architecture and users.
My background combines statistics, data science, machine learning, computer vision, AI systems, data and cloud systems, and hands-on engineering. Rather than defining myself by one technology, I start from the problem, make the trade-offs explicit and choose the approach that fits.
I enjoy staying hands-on—investigating data, prototyping and building critical components—while also shaping interfaces, scalability, reliability and the path beyond the first demo. I work best where technical depth, cross-functional collaboration and operational constraints meet.
Spanish · English · French
08 / Let’s connect
A complex Data & AI problem?
Let’s find a way forward.
For conversations about data, AI, systems engineering and intelligent automation.
Get in touch