Bryan Javier Duarte

AI Engineer with 5+ years of software engineering experience (5+ YOE), including 3+ years as a Tech Lead driving technical roadmaps and 2+ years delivering production-grade AI solutions. Dual AWS certified (AI Practitioner and Cloud Practitioner) with strong full-stack foundations.

Specialized in architecting and operating production-grade AI systems, focusing on Agentic AI, orchestration with LangChain/LangGraph, and deep observability.

I build resilient software by integrating AI capabilities through context management, tool use, advanced RAG, and comprehensive monitoring. I combine a product-oriented mindset with deep technical rigor to deliver end-to-end AI solutions: from initial prototyping and development to production deployment and post-launch monitoring.

AI Engineering & LLM Ops

Agentic AI Systems, LangChain, LangGraph, LangChain Deep Agents, RAG (Retrieval-Augmented Generation), Prompt Engineering, Hallucination Monitoring (LangSmith/Langfuse), Embeddings, Vector Search

Fullstack Development

Python (Advanced), TypeScript, Node.js, Next.js, React, API Integration, Microservices, RESTful Services

Cloud & Infrastructure

AWS (Bedrock, Serverless, Lambda, EC2, S3, ECS, ECR, CloudWatch, SAM, RDS, App Runner, Amplify), Terraform (IaC), Docker, PostgreSQL, PgVector, Redis, CI/CD (GitHub Actions), CodeArtifact

AI Development Tools

MCPs, Cursor | Advanced Claude Code (proficient in all primitives and the Claude Agent SDK to embed into CI/CD pipelines and automations; authored custom Claude Code Plugin)

Other

Self-hosted n8n, Sentry, Advanced Excel

Panóptico AI

Production-grade autonomous news intelligence platform that clusters and analyzes event feeds using agents, embeddings, and AI workflows. Read Technical Blog Post

1. Vera (Fintech)

   United States (Remote)

   AI EngineerApril 2026 – Present · 6 months

   • Transitioned the agent's core from a centralized single-router architecture to an orchestrated network of subagents using LangChain Deep Agents, specializing domain handling (financial and operational) and reducing user intent drift.
   • Designed and implemented the user long-term memory architecture (personal and financial context), managing asynchronous ingestion and prioritization via a dedicated attention-queue pipeline.
   • Re-engineered the voice agent over WebSockets and WebRTC streaming, integrating an in-flight memory layer and low-latency RAG for real-time conversational responses.
   • Engineered an internal, CLI-driven LLM evaluation framework integrated into CI/CD pipelines, drastically reducing hallucinations and standardizing production observability with Langfuse.

2. FirstJob

   Santiago, Chile

   Tech Lead (Promotion)July 2022 – March 2026 · 3 years 8 months

   • Architected and deployed an autonomous agentic system using LangGraph and LangChain, featuring multi-step reasoning (autonomous tool-calling), long-term memory management, and cyclical clustering/refinement pipelines.
   • Built a hybrid RAG engine using PostgreSQL and PgVector (semantic vector search + full-text search), optimizing context window packing to reduce inference latency and compute costs.
   • Deployed continuous agent monitoring and tracing using LangSmith (later migrated to Langfuse), ensuring output auditing and proactive hallucination mitigation.
   • Significantly reduced AI operational expenditure (OpEx) and inference latency by implementing prompt caching, batch generation APIs, and dynamic multi-provider routing.
   • Developed content safety architectures based on the Generator-Validator pattern (Guardrails), enforcing strict structured outputs (JSON Schema / Pydantic) and distributed multi-provider rate limiting.
   • Defined team-wide AI engineering standards, leading cross-functional adoption of Claude Code and integrating Claude Agent SDK into CI/CD for automated PR code reviews and continuous synchronization of CLAUDE.md and AGENTS.md.
   • Automated 100% of the job post auditing workflow using an autonomous multi-agent system, eliminating human error and enforcing strict platform editorial compliance.
   • Architected and scaled both frontend and backend for an internal greenfield CRM, centralizing disparate product and operational databases to establish a single source of truth for Data and Product teams.
   • Led the architectural migration of mission-critical operational components to an AWS Serverless environment, optimizing infrastructure costs and accelerating agile deployments through CI/CD pipelines.
   1. Platform Project Leader (Promotion)April 2022 – July 2022
   2. Project Coordinator (Promotion)July 2021 – April 2022
   3. People Data & Ops TraineeJanuary 2021 – June 2021

• AWS Certified AI Practitioner

  Amazon Web Services

• AWS Certified Cloud Practitioner

  Amazon Web Services

• IBM AI Developer Specialization

  IBM

• AI Engineering & Full Stack Development

  Continuous Learning — Ongoing study of scalable architectures, LLM systems, and modern software design patterns

• B.S. in Certified Public Accounting & Auditing (Contador Auditor)

  INACAP Technological University of Chile

Spanish

Native

English

Professional Working Proficiency (B2 — Fluent technical communication)