About
Santiago Paz
I’m Santiago Paz, a software engineer with 13 years building high-traffic web platforms, now moving into AI engineering — retrieval, agents and multi-agent systems. I’m based in Berlin and EU work-authorized as an Italian citizen. Taking LLMs into production means the retrieval, agents, guardrails and evaluation around the model, not just an API call.
Most of my work is shipping real AI products, not demos: Contract Lens, a multi-tenant SaaS whose LLM extraction pipeline turns PDF and DOCX contracts into typed, structured data driving a deadline and alert system; Multi-Agent Trading Desk, 12 LLM analyst agents on a LangGraph + FastAPI backend driving rebalancing and backtesting over a portfolio; plus Reema, a text-to-speech SaaS, and Pegala, an AI job-search platform. I care about the parts that make AI features actually ship — evals, reliability, latency, and failure modes — as much as the model itself.
Alongside the products I build the fundamentals in the open: bedrock-genai-labsis 26 runnable labs against live Amazon Bedrock APIs — a FAISS vector store written from scratch, hybrid dense + BM25 retrieval with reranking, an agent loop implemented by hand before rebuilding it on the Strands Agents SDK, guardrails and PII redaction, and LLM-as-judge evaluation. It backs the AWS Certified Generative AI Developer — Professional certification I’m working toward.
Running underneath the last decade is a recurring thread: visual page builders. A WYSIWYG builder in Next.js at Dialpad, a low-code drag-and-drop builder in Angular at Monks, and a campaign page builder at R/GA — content models turned into systems non-engineers can drive.
On the stack: Python and TypeScript, with LangGraph, the Strands Agents SDK, FastAPI, FAISS and boto3 on the AI side, and React and Next.js on the front end — plus retrieval-augmented generation, guardrails and evaluation harnesses on Amazon Bedrock.
Before Berlin I was in Buenos Aires — I wrote about the move in Leaving Buenos Aires. I’m currently open to senior AI / LLM engineering roles.
At a glance
- Role
- AI Engineer
- Based
- Berlin, Germany
- Authorization
- EU work authorization (Italian citizen)
- Languages
- Spanish (Native) · English (Bilingual) · German (Elementary)
- Focus
- Production AI/LLM features · full-stack product delivery
- Status
- Open to senior AI engineering roles
Experience
13 years shipping production software — most of it on high-traffic web platforms — now focused on taking LLMs into production.
Build and ship pages across dialpad.com — a Next.js App Router site on Vercel with a Contentful back end, ~1,900 pages across five locales, serving millions of visitors. Currently building a WYSIWYG page builder in Next.js: pages composed visually from modular blocks, with each block's properties mapped and kept in sync with its Contentful content model. TypeScript, React and Tailwind, with performance, SEO and accessibility as release criteria.
Built and maintained high-traffic marketing pages across Dialpad's global web platform.
High-performance web solutions for major digital platforms — a low-code drag-and-drop page builder in Angular for mobile storefronts, and web showcases for Google Chromecast focused on seamless interactivity across devices.
Marketing pages for Google's Small Business initiatives (smallbusiness.withgoogle.com) in AngularJS, reaching millions of visitors monthly, plus a WYSIWYG page builder that let the team assemble campaign pages without engineering time. Held performance and WCAG compliance across high-traffic pages.
Built the platform behind Wrap (wrap.co), which generates "Wraps": app-like interactive content that runs inside messaging feeds with no app install. MEAN stack — Angular front end on Node.js services — with user, team and permission management, and a modular architecture so new content formats could be added independently of the generator.
Co-engineered JPMorgan's migration from a legacy ActionScript system to a full Angular platform, rebuilding the stock calculation interfaces used by brokers on top of the bank's existing back end, and refactoring interfaces for WCAG compliance.
FinTech and PropTech modules for the Swiss financial sector: financial planning, tax and real estate, in Angular and TypeScript.
Promotional rewards platform for GM's "GM Card" program in the US market, built in AngularJS.
Large-scale telecom workflows for HP's enterprise division, in J2SE, JBoss Seam and JSF.
Education & certifications
- Mar 2023
- Bachelor's Degree in Data Science — Universidad de Buenos Aires
- 2018 — 2019
- Data Science Program — Digital House
- 2010 — 2013
- Bachelor of Computer Science (incomplete) — Universidad de Buenos Aires
- In progress
- AWS Certified Generative AI Developer — Professional
Recognition
- 2026
- 3rd place globally — Hack-Nation Global AI Hackathon (2,000+ participants) — YETI Fellowship · Startup Factory boOst, Dresden
Frequently asked
Who is Santiago Paz?
Santiago Paz is an AI engineer based in Berlin, Germany, with 13 years of software engineering experience building high-traffic web platforms. He works on RAG, agents and multi-agent systems — the retrieval, guardrails and evaluation around the model, not just the API call — and is EU work-authorized as an Italian citizen.
What does Santiago Paz build?
He ships production AI/LLM systems end-to-end: Contract Lens, a multi-tenant SaaS whose LLM extraction pipeline turns PDF and DOCX contracts into typed, structured data driving a deadline and alert system; Multi-Agent Trading Desk, 12 LLM analyst agents on a LangGraph + FastAPI backend; and bedrock-genai-labs, 26 runnable Amazon Bedrock labs covering RAG, agents, guardrails and LLM-as-judge evaluation. Stack: Python and FastAPI on the back end, TypeScript, React and Next.js on the front end.
Where is Santiago Paz based?
Berlin, Germany. He is EU work-authorized as an Italian citizen and open to senior AI engineering roles across the EU and remotely.
What is Santiago Paz's tech stack?
Python and TypeScript, with LangGraph, the Strands Agents SDK, FastAPI, FAISS and boto3 on the AI side, and React and Next.js on the front end. On AWS he works with Amazon Bedrock — retrieval-augmented generation (RAG), agents and tool use, guardrails, and evaluation harnesses that catch quality drift.
How can I contact or hire Santiago Paz?
Email santiago.paz.1992@gmail.com, or reach him on GitHub (github.com/santiago-paz) and LinkedIn. He is open to senior AI/LLM engineering roles.
Elsewhere
- santiago.paz.1992@gmail.com
- GitHub
- github.com/santiago-paz
- linkedin.com/in/santiago-paz