AI ENGINEERING · ALICANTE / EN + ES
Documents that become data. Knowledge your team can find. Images and speech that help you decide.
Three engineers. A focused pilot. Evidence to decide what comes next.
Drag to rotate. Once focused, scroll to zoom. Double-click to reset. Keyboard: arrows to rotate, + or − to zoom, Home to reset, P to pause.
01 / SELECTED WORK
Six demonstrations to help you imagine your next tool. See how information enters, how it is reviewed and what each workflow delivers.
Explore demo
Extract → review → deliver
From a scanned invoice to an approved record in another system.
81 of 84 cases · Fictional invoicesFields correctly extracted before review
Explore demo
Compatible photographs and three landmarks to explore in 3D.
39 of 39 photos · Attributed public photographsPhotos registered across three independent models
Explore demo
Ask → retrieve → verify
Search your documents and see the source behind each answer.
16 of 16 cases · synthetic dataQueries with the expected source and page
Explore demo
Compare → locate → review
Inspect real circuit boards and review anomalies and errors.
76 of 80 cases · Real images · VisAHeld-out images correctly classified
Explore demo
Dictate → structure → export
Turn scattered notes into reports you can review.
24 of 24 cases · synthetic dataCases with expected fact coverage
Explore demo
Transcribe → review → summarise
Transcribe recordings and review their key information.
12 of 12 cases · synthetic dataCases with expected fact coverage
Independent demos with identified public or fictional inputs. Every case includes its evaluation and limitations.
02 / WHAT WE CAN BUILD
We connect models, software and human review around a specific team workflow.
Extraction, validation and reconciliation with human review. From an invoice to an approved record in the receiving system.
Search manuals and internal documentation with visible sources. We define the tools and permissions each assistant needs.
Prepare reports and conversation summaries your team can review, correct and export in an agreed format.
We also work with inspection, image matching, reconstruction and annotation.
See specialist work →03 / A PRACTICAL FIRST STEP
Find out how AI fits your workflow before committing to a larger build. Start with a data sample and a question worth answering.
Scope my pilot ↗A focused first milestone · Budget and timeline scoped together
WHAT YOU RECEIVE
Scope, deliverables and acceptance criteria agreed before building.
An interface or API, usage examples and an evaluation on an agreed sample.
Results, errors, limitations and a recommendation for the next step.
“Extract the agreed fields, flag uncertain data and export only what a person approves.”
HOW WE WORK
A clear point of contact and technical review between teammates. We choose models for quality, cost, privacy and project requirements.
Review the workflow, a sample and constraints. Identify what is worth automating.
Define a first milestone, its cost, deliverables and how to check the outcome.
Test the workflow, show its errors and deliver a reviewable demo or API.
Use the results to decide whether to expand, integrate or rethink the approach.
04 / THE TEAM
We are Alejandro, Ander and Sergio. We combine AI, robotics and software development to turn an idea into a tool you can use and evaluate.
Applied AI · RAG · Research
AI engineer focused on RAG and machine learning. Combines document retrieval, generative AI and research to build solutions from data.
Computer vision · OCR · Document AI
AI engineer focused on computer vision and document processing, from detection and segmentation to data extraction, review and integration.
Audio · Multi-agent systems · Local models
AI engineer focused on audio, agents and local models. Designs transcription, analysis and speech synthesis systems with coordinated tools and private inference.
TEAM BACKGROUND
2026 / IEEE ICRAPublication by Alejandro Climent Peñalver ↗
2025 / Sopra SteriaNational winners with Professor Who · Alejandro, Ander and Sergio ↗
Cervera 5RNational instance-segmentation challenge · Ander and Sergio ↗
2024 / SolarIACódigo Solar winners · Alejandro Climent and Alejandro López Sánchez ↗Sensor-based predictive-maintenance proposal. A prior project by its authors; no technical evaluation published in this portfolio.
05 / YOUR NEXT STEP
Tell us what you do today, what data you work with and what outcome would help. We can use that to scope a useful proof of concept.
Talk to Quarkeon on Upwork ↗Alicante, Spain · Working in English and Spanish
