Skip to content
View santos-sanz's full-sized avatar
🎯
Focusing
🎯
Focusing
14 followers · 64 following

Achievements

Achievement: Pull Sharkx3Achievement: QuickdrawAchievement: YOLO

Achievements

Achievement: Pull Sharkx3Achievement: QuickdrawAchievement: YOLO

Highlights

  • Pro

Block or report santos-sanz

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
santos-sanz/README.md
Andrés Santos Sanz — Applied AI, workflows and automation

Applied AI Lead @ Bit2Me · ex-Revolut · Based in Spain

I turn AI, data, and automation into practical systems that help teams
make better decisions, remove repetitive work, and ship reliable workflows.

LinkedIn  ·  Email

Building AI that works beyond the demo

My work sits at the intersection of applied AI, product judgment, data, and operations. I focus on useful systems: tools that fit real workflows, make their constraints explicit, and can be reviewed and verified by the people who depend on them.

Selected work

Project What it demonstrates
Lifeskills
AI for practical business work		 Reusable agent skills for communication, strategy, negotiation, and influence.
AI Dev Workflow
Agentic software delivery		 A practical operating system for building software with Codex, Claude Code, OpenCode, and Cursor.
Agent Guardrails Template
Reliability by design		 Guardrails, quality gates, and executable verification loops for dependable agent workflows.
MLX LoRA Fine-tune Template
Open-source technical depth		 An end-to-end workflow for fine-tuning LLMs with LoRA on Apple Silicon.
Doom Slayer
Private, on-device AI		 A browser-based focus assistant using MediaPipe and TensorFlow.js with no backend.
DataClaw
Agents for data work		 A Kaggle query agent that runs SQL on DuckDB and falls back to Python when needed.

How I build

  • Outcome first: start with a workflow, decision, or user problem—not a model.
  • Human in the loop: increase leverage while keeping review and accountability.
  • Reliable by design: use explicit constraints, quality gates, and verification loops.
  • Proof over claims: prefer small working artifacts to abstract opinions.

Focus

AI agents · workflow automation · LLM applications · data & operations
product experimentation · AI-assisted development · agent reliability

Let's connect

I am always interested in thoughtful conversations about applied AI, reliable agent workflows, product experimentation, and automation for teams.

Connect on LinkedIn · Send me an email

Pinned Loading

  1. lifeskills lifeskills Public

    Non-coding skills for AI agents, focusing on high-impact skills.

    Python 3

  2. ai-dev-workflow ai-dev-workflow Public

    AI-assisted development workflow pack for Codex, Claude Code, OpenCode, and Cursor with native-first skills, OpenSpec, and review/ship guidance.

    Python

  3. agent-guardrails-template agent-guardrails-template Public

    Template for agent guardrails, AGENTS.md workflow, quality gates, and executable verification loops.

    Shell

  4. mlx-lora-finetune-template mlx-lora-finetune-template Public

    Template for fine-tuning LLMs with LoRA using Apple MLX on Mac Silicon

    Python 8 1

  5. doom-slayer doom-slayer Public

    AI-powered focus assistant using MediaPipe and TensorFlow.js to detect distraction (phone use, looking away) and keep you productive. 100% local, no backend.

    JavaScript 1

  6. dataclaw dataclaw Public

    Query agent for Kaggle datasets with SQL execution on DuckDB and Python fallback when needed.

    TypeScript