AI Architect & Technical Director/Agentic & LLM Systems in Production/Stanford Top 2% Scientist

Jose L.
Salmeron, PhD

◆ Technical Authority
Directs and builds agentic AI

I own the technical direction — multi-agent architectures, private LLMs, knowledge graphs, RAG — with the depth to be accountable for what ships. Today: directing and building agentic systems at the frontier.

● Leader & Scientist
With the science & scale to lead it

€15M+ programmes, €20M+ secured as PI, multidisciplinary teams to 60. Proven by science: Full Professor (Catedrático), h-index 39, 200+ peer-reviewed papers.

A rare technical leader who directs and builds: I own the architecture of agentic and LLM systems in regulated production — and stay deep enough in the engineering to be accountable for whether they ship. And I have led the €15M programmes, multidisciplinary teams and grant-winning science that scale them. Sharpest exactly where agentic AI gets hard: agent governance, responsible AI and reliable deployment.

Work with me → See what I build →
Jose L. Salmeron, PhD
Current
Director of Agentic AI
@ Accenture
Agentic AIArchitecture & production deployment
Top 2%Stanford Global Scientist Ranking
€20M+Total research funding secured as PI
$ stack --core Agentic frameworks Agent interop vLLM llama.cpp Knowledge Graphs RAG Quantization Federated Learning XAI Python · PyTorch
Scientific authority
Top 2%
Stanford Ranking
h · 39
H-Index
7,000+
Citations
200+
Publications
Leadership & delivery
€20M+
Secured as PI
€15M
Largest programme led
60
Team led
20 yrs
In enterprise AI
01 — What I Build

Technical authority at the frontier of agentic AI

I own the technical direction of agentic and LLM systems and build them — multi-agent orchestration, private LLM deployment, and the grounding, observability and governance that make them dependable in regulated production. Directing and engineering at once is rare, and it is precisely where agentic AI succeeds or fails.

01 / agentic-architecture

Multi-agent architectures

Design and production-grade Python to deploy and orchestrate agents at enterprise scale — currently accelerating SAP migrations with agentic AI. Scalable, robust, and built to be governed.

agentic frameworksorchestratorsmulti-agent patternsinterop protocols
02 / private-llms

Private & efficient LLMs

Private, on-prem and efficient LLM deployment where data can't leave the building — fine-tuning, quantization and model compression for cost-efficient inference at scale.

vLLMllama.cppOllamaLlamaquantization
03 / grounding

Grounding & reasoning

Ontologies and knowledge graphs as the backbone of agent grounding and reasoning; advanced RAG, memory and retrieval so agents act on truth, not hallucination.

Knowledge GraphsOntologiesRAGvector DBs
04 / evals-and-cost

Observability & cost control

Output-quality metrics and evals for agents, plus token-saving strategies — per-task model routing, private/small models where they fit, prompt optimization — to keep quality up and spend down.

evalsobservabilitymodel routingAgentOps
Open source — designed, written and released solo
swarmstate Rust · Python →
Framework-agnostic state management and checkpointing backend for agent orchestration — the plumbing that keeps multi-agent runs recoverable.
TinyLlama 1.1B · Mac GGUF Hugging Face →
My own quantization of TinyLlama-1.1B-Chat (~700 MB) for Apple Silicon — solo work on model compression for efficient on-device inference.
TurboSwarm Rust · Python →
Full-featured particle swarm optimization library — metaheuristics for hard search and optimization problems.
fuzzytool Python · NumPy →
Fuzzy-logic toolkit for building fuzzy inference systems — building on three decades of published work in Fuzzy Cognitive Maps.
GraphFinder → GitHub → Hugging Face →
02 — Track Record

Built, led, and delivered at scale

Twenty years taking AI from strategy to production in highly regulated, high-stakes environments — banking, insurance, healthcare, industrial — where governance and production-grade delivery are non-negotiable.

Sectors

Insurance · Banking · Healthcare · Industrial · Public sector. Wherever getting AI right is non-negotiable and audit-ready delivery matters.

Accenture

Director of Agentic AI

Technical lead for agentic AI on an SAP migration-acceleration programme. Deliberately hands-on: agent architectures, ontologies/knowledge graphs, observability and inference-cost optimization at scale.

2026 –
AI Venture · Stealth

Chief AI Officer

Owned end-to-end AI strategy and product. Built private enterprise LLMs and distributed training for NLP; senior GenAI advisor to a global insurer (MAPFRE Global Corporate).

2025 – 2026
Capgemini · Hybrid Intelligence

Principal AI Director / Tech Director

Directed a €15M NextGenEU programme (federated learning, XAI, deep learning) with teams to 60 in a regulated healthcare setting. Technical Authorities Board member.

2019 – 2025
BBVA · New Digital Business

Lead Data Scientist

Led AI/ML in a top-tier regulated bank: explainable, audit-ready, bias-mitigated risk-scoring and portfolio models. Secure multiparty computation, data-poisoning detection.

2018 – 2020
European Commission · FP7

Director, Technical Committee

Directed the technical committee of a multi-stakeholder intelligent resource-management programme — real-time prediction, optimization, multi-source data — owning roadmap and governance.

2014 – 2017
Enterprise proposals

Technical win owner

Led the technical side of winning enterprise RFPs — SEAT-CUPRA (awarded) and MASORANGE — shaping the value case and solution architecture that secured the business.

SEAT-CUPRA · awarded
03 — Flagship Programmes

From architecture to production

At a glance

A selection from €20M+ in funded R&D and enterprise programmes led as principal investigator or technical director — from agentic systems to federated healthcare AI.

€2M
2026 –
Accenture · SAP

RSA — Rapid SAP Adoption

Agentic AI programme to accelerate enterprise SAP migrations. I own the ontology and knowledge-graph design that grounds the system, and the agentic architecture built around it — the reasoning and orchestration layer that lets agents act reliably on that structured knowledge in production.

Agentic AIOntologyKnowledge GraphsSAP
€15M
2021–2024
NextGenEU · Capgemini

AI4HealthyAging — Distributed AI for Early Disease Diagnosis

As principal investigator, I led the architecture and delivery of a federated, explainable AI platform enabling early diagnosis of age-related diseases across distributed hospital networks — without centralising patient data. Cross-European consortium, inception to production, regulated setting.

Federated LearningHealthcare AIXAIDistributed Systems
$250K
2024–2026
Canadian CIHR · Digital Twins

CARDIO-TWIN — Cardiovascular Risk Detection in Women

I lead the technical and research work on digital-twin AI for early cardiovascular risk from scarce clinical data — demonstrating that explainability and predictive performance coexist even in data-constrained clinical environments. Algorithm design and research leadership.

Digital TwinsClinical AIXAI
Aerospace
2020
Airbus Defence & Space · Military

A400M Wing-Fit — ML Virtual Photogrammetry for Airbus Military

I led the feasibility, algorithm design and experimental work for a machine-learning-based virtual photogrammetry system to automate wing-fit on the Airbus A400M military transport — 3D computer vision applied to a high-precision aerospace manufacturing problem.

Computer Vision3D / PhotogrammetryAerospaceMachine Learning
Prod
2018–2020
BBVA New Digital Business

Explainable AI for Financial Decision Systems at Scale

I designed and deployed XAI risk-scoring and portfolio-optimization systems for audit-ready financial decisions — meeting regulatory transparency requirements within a single unified architecture, using secure multiparty computation.

XAIFinanceRegulatory ComplianceSMPC
04 — Research

The science underneath

Double PhD
AI & Management Info. Systems · both Summa Cum Laude · + MBA
4×
National research sexenios · Spanish Ministry of Science
20+
Countries citing the work
Award
PhD Extraordinary Dissertation Award
Editorial & Program Chairs
Program Chair — 18th Intl. KES Conference on Intelligent Decision Technologies, Barcelona, 2026
Area Editor — Evolving Systems (Springer Nature)
Editorial Board — Mathematics (Q1 JCR)
Vice-President — Intl. Association of Grey Systems & Uncertain Analysis
2026
Neurocomputing
Convergence and Stability in Algebraic and Exponential Fuzzy Grey Cognitive Maps
Salmeron & Pueyo
2026
Neurocomputing
Divergence-Driven Generative Adversarial Networks for semi-supervised anomaly detection
Garcia & Salmeron
2026
IEEE Transactions on Fuzzy Systems
Non-monotonic Causal Discovery with Kolmogorov–Arnold Fuzzy Cognitive Maps
Salmeron
2026
Knowledge-Based Systems
Spectral Biclustering-Driven Scalability for Post-Hoc Explainability in Recommender Systems
Salmeron & Arevalo
2026
Mathematics
Pharmacokinetics-Informed Agentic Architecture for Drug Dynamics with LLM-Driven In Silico Patients
Salmeron
Google Scholar → ORCID → Academic profile →
Built & delivered with
Accenture Capgemini BBVA MAPFRE Intel Airbus Vodafone Cisco Gilead
05 — Work with me

Where I can help

Availability

Based in Madrid · open to remote (EU / US time zones). Selective engagements.

I partner with organisations that need to take agentic AI and LLM systems from prototype to reliable, governed production — especially under EU AI Act and regulatory constraints. That can be a hands-on technical lead role, a Head/VP of AI mandate, senior advisory, or a funded research collaboration. If you're scaling AI where getting it right matters, let's talk.

Emailsalmeron@acm.org→ LinkedInlinkedin.com/in/salmeron→ GitHubgithub.com/joselsalmeron→ ScholarGoogle Scholar Profile→
Engagement areas
Agentic AI & LLM architecture (hands-on)
Head / VP of AI · Technical Director
Private LLM & RAG systems in production
AI governance & EU AI Act readiness
XAI audits for regulated industries
Board-level & C-suite AI advisory
EU / national R&D programme leadership
Keynotes & executive AI briefings