Marcos García Estévez

AI / GenAI Engineer | LLMs, RAG, AI Agents & Applied AI

Málaga, Spain [email protected] linkedin.com/in/marcosgarest github.com/warc0s

Professional profile

Applied AI Engineer specializing in Large Language Model (LLM), Retrieval-Augmented Generation (RAG) and agentic systems. Experience on client projects involving LLMs and generative AI at Accenture, alongside self-built products and tools shipped to production and open source. Strong Python, backend and NLP/ML foundation, with regular use of Claude Code and other AI-assisted development environments.

Professional experience

Full Stack LLM Development Analyst - Accenture Spain

Jun 2025 - Present

Málaga, Spain · Hybrid

Contributing to client projects that incorporate LLMs and generative AI, taking part in the development and implementation of different solutions.
Intensive daily use of Claude Code alongside other AI-assisted development environments for analysis, prototyping and implementation.
Development and integration of software components within client-facing teams, working mainly with Python, APIs and cloud services.

Full Stack LLM Intern - Accenture Spain

Mar 2025 - May 2025

Fine-tuned and evaluated BERT models for multilingual classification; automated model comparisons and tracked experiments with MLflow.

LLM Quality Specialist - Outlier

May 2023 - Jul 2024

Remote · Freelance

Systematically evaluated Spanish LLM responses for factual accuracy, relevance, instruction following, reasoning, inconsistencies and hallucinations.
Reviewed training/evaluation data and model outputs, providing structured feedback to improve model quality and usefulness.

Selected projects

Awaita (Awi)awaita.tech

Production-oriented AI guide for Menorca. Work on the conversational layer: RAG over controlled sources, retrieval and reranking quality, evaluation, safety/abuse controls, feedback, latency/cost/quality observability and staged CI/CD releases.

SoloGangassologangas.es

Production deal platform with a public website and a 2.3K+ Telegram audience. End-to-end automation with AI-assisted processing, multi-channel publishing and lifecycle maintenance; Python/PostgreSQL backend, Docker operations, monitoring, backups and staged CI/CD with health checks and rollback.

OpenCode Permission Reviewergithub.com/warc0s/opencode-permission-reviewer

Automatic permission-review plugin for coding agents: tool-free LLM reviewer, configurable policy, deterministic risk invariants, secret redaction, audit logging, read-only evidence enrichment and fail-safe escalation to human review.

Technical skills

LLM / GenAI: RAG, hybrid retrieval, embeddings, reranking, prompt design, structured outputs, tool calling, agents, LLM evaluation, MCP.

Frameworks / APIs: LangChain, LangGraph, LlamaIndex, FastMCP, OpenAI API, Gemini, Transformers.

Backend / Cloud: Python, FastAPI, Flask, REST APIs, SQL, Docker, AWS Lambda, Linux, CI/CD, observability.

ML / Data: BERT, scikit-learn, MLflow, TensorFlow/Keras, Pandas, DuckDB, SQLite, vector search.

Education & certifications

Advanced Vocational Specialization in Artificial Intelligence & Big Data · CPIFP Alan Turing - Accenture
600-hour dual program · Final grade: 10/10 · Capstone: MIDAS

2024 - 2025

Higher Vocational Diploma in Web Application Development · MEDAC

2022 - 2024

Selected certifications: GitHub Copilot (Microsoft, 2026) · AWS Academy Cloud Foundations · PCEP (Python Institute).

Languages: Spanish - Native · English - B1 (Cambridge).