marisame_

// curriculum_vitae

Raul Vasquez

Machine Learning Engineer · AI Systems

// summary

Machine Learning Engineer with solid experience in AI, data engineering, and cloud computing. Specialized in designing, optimizing, and deploying scalable ML systems and full-stack applications. Builds advanced agent-based architectures and retrieval-augmented generation (RAG) systems with LangChain/LangGraph, integrating tool orchestration, persistent memory, and human-in-the-loop workflows.

// skills

AI Agents & Orchestration

LangGraph (checkpointer, store, HITL middleware)LangChain agents (deepagents, create_agent)Tool orchestration & multi-step reasoningStreaming token responses (SSE / WebSocket)

RAG & Vector Search

PGvector, ChromadbHugging Face & Ollama embeddingsHybrid retrievalDocument chunking & ingestion

MCP (Model Context Protocol)

MCP servers (fastapi-mcp, FastMCP)MCP clients (MultiServerMCPClient)MCP Oauth & API key authentication

Backend & Data

FastAPISQLAlchemy async + asyncpgAlembicPostgreSQL, SupabaseRedisJWT + refresh tokensPolymorphic RBACMulti-tenant architecture

ML & Deep Learning

TensorFlowPyTorchScikit-learnKerasVision-Language models (X-CLIP)LLM fine-tuning (LLaMA)Prompt engineering

Frontend

ReactNext.js (App Router)React Router v7TailwindCSSTypeScript

Cloud & DevOps

AWSDockerKubernetesCI/CD (Python + Bash)

Languages & Tools

PythonTypeScriptGo (GraphQL)SQLuvpnpmGit

// experience

Full-Stack Developer & AI Engineer

2025 – Present

W3bInnovation · Remote

  • Designed and deployed intelligent-agent architectures with LangChain & LangGraph — agent workflows, tool orchestration, persistent memory (PostgreSQL checkpointer + store), and structured decision pipelines.
  • Built multi-tenant RAG systems with pgvector and Ollama embeddings, improving factual consistency and traceability of conversational AI for enterprise clients.
  • Implemented an MCP server (fastapi-mcp) exposing project, document, and RAG tools to external agents — with API-key authentication, scoped permissions, and request logging.
  • Developed FastAPI services for LLM inference, authentication, session management, and agent lifecycle orchestration; designed a polymorphic RBAC system with resource-grant inheritance (organization → project → conversation).
  • Built React + TailwindCSS + React Router v7 frontends with WebSocket streaming for real-time agent token responses, tool-use indicators, and human-in-the-loop interrupts.

Machine Learning Engineer

2022 – 2024

Datyra · San Diego, USA

ML & Data Modeling

  • Developed and deployed scalable ML systems with TensorFlow, PyTorch, Scikit-learn, and Keras: regression, classification, clustering.
  • Fine-tuned vision-language and LLM models (X-CLIP, LLaMA) for object detection, semantic understanding, and domain-specific response generation.
  • Built LLM-powered systems for automated reporting, structured data analysis, and conversational interfaces using prompt engineering and customized RAG pipelines.

Cloud & Data

  • Managed PostgreSQL, MySQL, MongoDB; designed scalable architectures on AWS and Azure.

Automation & Deployment

  • Designed automated training/validation/deployment pipelines (Python + Bash, CI/CD); deployed containerized apps via Docker and Kubernetes.

Software & Web

  • Implemented GraphQL services in Go; integrated Stripe payments and referral systems; built secure RBAC authentication.

Productivity & Visualization

  • Built dashboards and internal tools with Apache Superset, Mercury, and Retool; managed audience segmentation via Mailchimp.

// projects

This monorepo — Next.js 14 + FastAPI + LangGraph + pgvector. Live demos for streaming chat, RAG, and agent visualization. Provider-agnostic LLM (Ollama in dev, OpenAI in prod).

Next.jsFastAPILangGraphpgvectorTurborepo
Lyra

Python CLI agent with MCP tool integration, persistent memory (filesystem + PostgreSQL), configurable models (Ollama local + cloud), and thread management.

PythonDeepAgentsMCPOllamaPostgreSQL
mcp-servers

Chat platform on LangGraph + Chainlit + MCP. Uses AsyncPostgresSaver/Store, summarization / HITL / todo-list middleware, and multi-server MCP integration.

LangGraphChainlitMCPOllamaPostgreSQL
Enterprise RAG Platform

Client project (anonymized). FastAPI backend with LangGraph SDK; React Router v7 + Tailwind v4 frontend. Polymorphic RBAC, MCP server with API keys, conversation sharing, WebSocket token streaming.

FastAPILangGraph SDKReact Router v7pgvectorfastapi-mcp

// education

Master in Robotics

Universidad Tecnológica de la Mixteca

2019 – 2021

Bachelor in Mechatronics

Universidad Tecnológica de la Mixteca

2014 – 2019

// languages

  • SpanishNative
  • EnglishProficient

// certifications

  • Applied Data Science with Python Specialization (Mar 2022)
  • SQL and PostgreSQL: The Complete Developer's Guide (Feb 2022)
  • IBM Data Engineering Specialization (Feb 2022)
  • NoSQL, Big Data, and Spark Foundations Specialization (Apr 2022)
  • BI Foundations with SQL, ETL, and Data Warehousing Specialization (Apr 2022)
  • Python for Everybody Specialization (Feb 2022)