Gonzalo Martin PerezAI Engineer | Software Engineer | Product Engineer

CV

Gonzalo Martin Perez

AI Engineer | Software Engineer | Product Engineer

LLM Agents · Agentic AI · RAG · GraphRAG · LLM Evals

Bahía Blanca, Argentina · 100% Remote (UTC-3) · English B2

PDF · 70.9 KiB · 2 pages

Professional Summary

AI Engineer with 2.5+ years shipping production software, most recently LLM agents, RAG systems and eval suites. Lead engineering at an early-stage fintech and blockchain startup whose generative interface opens digital assets to everyday people, and own its AI assistant end to end: the LangGraph agents in English, Spanish and Portuguese, the eval suite that gates every release, and the backend, web and mobile app around them. Previously built a GraphRAG assistant for Payway's merchant platform.

Professional Experience

Rampy (opens in a new tab)

AI Engineer

Sep 2026 – Present · United States · Remote

Leading engineering at an early-stage fintech and blockchain startup whose non-custodial app opens digital assets to everyday people through a generative interface. Working closely with the three founders and with specialists in blockchain and integrations.

  • Built the eval suite that gates every release of a multilingual AI assistant: 1,000+ scenarios replicated in English, Spanish and Portuguese (3,000+ cases), with the agents passing 90%+. An LLM-as-a-judge in LangSmith, validated against human review, grades every run.
  • Migrated the assistant from Agno to a multi-agent LangGraph system with cycles, financial tools, background workers, retries and failure recovery; Claude is the primary model, with DeepInfra and Vertex AI fallbacks, GraphRAG (Neo4j, pgvector) and Mem0 memory.
  • Made agent responses 2× faster and cut tokens by 40% by removing redundant tool calls and tightening memory and context, verified with load tests and evals.
  • Designed the structured-output contract that turns agent answers into generative UI in the React Native app: native buttons, cards and deep links into swap, vault and perps flows on live market data. Users confirm every transaction in a non-custodial Privy wallet.
  • Re-architected the inherited backend and frontend around Clean Architecture, cutting build-and-deploy from about 1 hour to 10 minutes. Built the trading integrations and fee model on top: Hyperliquid perps (200+ markets), LI.FI swaps (100+ tokens) and Morpho, Aave and Compound vaults, tracked via webhooks.

Teamcubation (opens in a new tab)

Software Engineer

Dec 2025 – Sep 2026 · Argentina · Remote

Consultant on Payway's merchant promotions platform (350,000+ merchants, 13 countries).

  • Built a GraphRAG assistant that passed 95%+ of a 500+ case eval suite, graded in LangSmith by a human-validated LLM-as-a-judge. Used LangGraph, Neo4j and PostgreSQL/pgvector, and added guardrails against off-topic questions and data leaks.
  • Delivered the event-driven pipeline that ingested 20M+ historical promotions into production (S3 and SQS into a Python/FastAPI Lambda, then Java/Spring Boot and NestJS services on Docker and Kubernetes). Wrote tooling to replay loads locally and reconcile failed records.

Cooperativa Obrera (opens in a new tab)

Software Engineer

Dec 2024 – Dec 2025 · Bahía Blanca, Argentina · On-site

Primary engineer on the permissions back office of Argentina's largest consumer cooperative (2.7M+ members).

  • Integrated 10+ internal systems into one admin experience through a Python/FastAPI BFF and a Next.js front end; a 10-minute manual permission lookup now takes under 10 seconds.
  • Cut permission tickets and manual assignments by 80%+ and recurring assignment bugs by 95%+ by standardizing how new systems are onboarded.

Self-employed

AI Engineer & Full-Stack Developer

Jan 2024 – Present · Argentina · Remote · Project-based

AI and full-stack projects for clients selling on Shopify and Hotmart; several are live in production.

  • Cut manual processes from hours or days to minutes with WhatsApp assistants, n8n, ManyChat and Meta Ads automations, and MCP content agents on the Claude or OpenAI APIs.
  • Delivered websites, custom software and e-commerce projects in days or weeks that generated USD 10,000+ in client profit, including Pequeverso (opens in a new tab)'s store and its GraphRAG shopping assistant on LangGraph.

Technical Skills

AI Engineering

AI Agents · Multi-Agent Systems · RAG · GraphRAG · LLM Evaluation · LLM-as-a-Judge · Tool Calling · Generative UI · Prompt Engineering · AI Guardrails

AI Frameworks & Platforms

LangGraph · LangChain · LangSmith (LLM Observability) · OpenTelemetry · Claude API · OpenAI API · Vertex AI · MCP · Neo4j · pgvector

Backend & Data

FastAPI · NestJS · Spring Boot · Laravel · REST APIs · Microservices · Clean Architecture · Event-Driven Architecture · PostgreSQL · Redis

Languages & Frontend

Python · TypeScript · Java · SQL · Node.js · React · Next.js · React Native · Kotlin · Swift

Cloud & Delivery

AWS Lambda · Amazon S3 · Amazon SQS · Docker · Kubernetes · GitHub Actions · SSE

Fintech & DeFi Integrations

Hyperliquid · LI.FI · Morpho · Aave · Compound · Privy · Non-Custodial Wallets

Selected Projects

Portfolio Assistant · Bilingual RAG Assistant with Evals (opens in a new tab)

Sep 2026 – Present

  • Open-source bilingual RAG assistant that answers recruiters' questions about my work: FastAPI and LangGraph with hybrid lexical and vector retrieval plus GraphRAG (pgvector, Neo4j), cited answers streamed over SSE and per-request cost limits. Implemented and evaluated; not yet deployed.
  • Built a 1,000-question bilingual eval bank plus adversarial cases, a 244-turn live-model evaluation on OpenAI and 800+ automated tests in GitHub Actions CI, with OpenTelemetry tracing.

Education