Available for new projects

Hi, I'm Diego De La Flor

~/

AI Developer · LLM Agents · RAG

I build LLM-agent skills and RAG pipelines that run in production and save engineering teams real hours.

View projects GitHub
3 Years of experience
70.9% Less time writing tests
5 Featured projects
3 Progressive roles

01. about

About me

Who I am, what I do, and the technologies I use to do it.

D

Diego De La Flor

AI Developer · LLM Agents · RAG

AI Developer with 3 years of experience and a QA → Full Stack → AI progression inside the same production HR tech. I am part of the team that maintains the engineering area's internal LLM-agent toolkit, and I own the QA vertical: I design, maintain and evolve every QA skill the team consumes. I designed the RAG pipeline that cut test-case creation time by 70.9% and refactored it into modular skills. My focus is maintainability: impact that holds up in production sprint after sprint, not just in a demo.

  • Location Lima, Peru
  • Education Software Engineering — UPC, Lima
  • Languages Spanish (native) · English (B1)
  • Availability Open to work — full-time / freelance

Courses & Certifications

  • Essentials Automation Certification Automation Anywhere · 2026
  • AWS Cloud Practitioner Essentials AWS Training · 2024
  • Excel Specialist IDAT · 2023
  • Cybersecurity Analyst UPC · 2022
  • SQL for Data Science UPC · 2022

Tech stack

AI & Agents
PythonRAGLLMsModel Context Protocol (MCP)Claude CodeOpenAI & Anthropic APIEmbeddingspgvector / ChromaDBPrompt & Context Engineering
Machine Learning
Random ForestXGBoostCatBoostscikit-learnIoT / ESP32
QA & Testing
PlaywrightCypressSeleniumPostmanApache JMeterBrowserStack
Backend
NestJSNode.jsSpring BootLaravelC# .NETREST APIsMicroservices
Frontend
ReactNext.jsAngularTypeScriptFlutterHTML5 / CSS3
Data & Cloud
PostgreSQLMySQLSQL ServerMongoDBFirebaseAzureAWSDockerGitHub ActionsJenkinsGit
Methodology
ScrumJiraConfluence

02. experience

Experience

Professional journey and the problems I have solved in production.

  1. Current

    AI Developer

    Mandü HR

    Feb 2026 – Present
    • Owner of the QA vertical inside the engineering team's internal LLM-agent toolkit: I design, maintain and evolve every QA skill the team consumes, within a shared, versioned platform distributed by role profiles.
    • Designed the RAG pipeline (Jira + Confluence, pgvector + ChromaDB) that cut test-case creation time by 70.9% — from 50h to 14.55h per sprint — and refactored it into reusable modular skills.
    • Built the skills for test-case generation, Jira upload, endpoint-to-Postman collection mapping, and Playwright E2E test generation with automatic evidence.
    • Keep the skills running in production: continuous maintenance, edge-case resolution and context optimization per skill, so the savings hold up sprint after sprint.
    • Iterate cross-team with the QA Tech Lead, my area's Tech Lead and the analysts on each team.
    PythonRAGMCPClaude CodeOpenAI APIAnthropic APIpgvectorChromaDBPlaywrightJiraPostman

  2. Full Stack Developer

    Mandü HR

    Jan 2025 – Jan 2026
    • Built 6+ high-impact production modules with Laravel, NestJS, Angular (Ant Design) and React + Vite, expanding the HR platform's functional coverage for a base of 4,500+ users.
    • Optimized critical SQL and NoSQL queries, reducing latency in high-concurrency read/write operations on the employee management system.
    • Proposed and implemented a refactor of the bulk-import pipeline using upsert with 10,000-record partitioning, achieving a 10 s reduction per batch and eliminating duplicates.
    • Integrated backend services with REST APIs in PHP/Laravel and NestJS within a microservices ecosystem, ensuring compatibility and code maintainability.
    LaravelPHPNestJSAngularReactViteTypeScriptSQLNoSQL

  3. QA Trainee

    Mandü HR

    Sep 2023 – Dec 2024
    • Ran functional and regression tests on web, mobile and desktop applications, documenting defects and validating requirements in agile sprints.
    • Automated API tests with Postman validating PHP endpoints, and ran load and stress tests with Apache JMeter, identifying bottlenecks and improving performance SLAs.
    • Validated compatibility and responsive design across multiple devices and browsers with BrowserStack, ensuring cross-platform coverage.
    PostmanApache JMeterBrowserStackAPI TestingPerformance TestingScrum

03. projects

Featured Projects

Systems I designed and built end-to-end, from architecture to production.

Public 1 project — code available
Private 4 projects — under NDA / proprietary
Private

QA Skills with LLM Agents — RAG + Playwright

Jan 2026 – Present

Problem

Manually writing test cases from user stories in Jira and Confluence consumed ~50 hours per sprint, creating QA bottlenecks. An isolated pipeline was not enough: the savings had to hold up in production and reach every team.

Architectural solution

RAG pipeline that ingests Jira and Confluence documentation (pgvector + ChromaDB, OpenAI/Anthropic API) and generates contextualized test cases. Refactored into modular skills inside the internal LLM-agent toolkit: test-case generation, automatic Jira upload, endpoint-to-Postman collection mapping (happy and unhappy paths), and Playwright E2E test generation that runs each case and captures evidence (screenshots and video).

Impact & Value

  • 70.9% reduction in test-case creation time (50h → 14.55h per sprint)
  • Skills in production, used daily by QA teams
  • Automated traceability: user story → test case → E2E test with evidence

Technologies

PythonRAGMCPOpenAI & Anthropic APIpgvectorChromaDBPlaywrightJira API
Private

DOM Capture & Test Generator

Feb 2026

Problem

Building E2E test scripts required manual, repetitive DOM inspection to extract robust selectors, slowing down test coverage for new features.

Architectural solution

Desktop application built with Electron, React, and Vite that visually captures the DOM state in real-time during user navigation. Automatically generates JSON component mappings (Page Object Model) and produces reusable E2E tests ready for Cypress.

Impact & Value

  • Radical acceleration in coding and maintaining E2E tests
  • Standardized selectors, reducing test suite fragility
  • Coverage for new features without manual DOM inspection

Technologies

ElectronReactViteTypeScriptCypressDOM Analysis
Private

Agricultural Recommendation System — Coffee

Aug 2024 – Dec 2025

Problem

Coffee farmers lack personalized, data-driven guidance to optimize yield and crop quality based on real environmental conditions.

Architectural solution

Dual backend system: Spring Boot for local IoT sensor data processing and .NET (Azure) for cloud logic. Data pipeline from ESP32 sensors (humidity, NPK, rain) to a classification model that generates contextual recommendations, exposed via REST API.

Impact & Value

  • 3 models evaluated (Random Forest, XGBoost, CatBoost) on ~9,983 samples; Random Forest selected for production
  • Real-time integration of 4 IoT sensor types
  • Dual local/cloud backend for offline resilience

Technologies

Spring Boot.NET / AzurePythonRandom ForestXGBoost / CatBoostESP32 / IoTSQL ServerREST APIs
Private

YouTube Shorts Content Automation

Feb 2026 – Apr 2026

Problem

Producing shorts consistently requires coordinating script, images, voice and publishing: hours of repetitive manual work per piece.

Architectural solution

End-to-end flow orchestrated with n8n that chains OpenAI (script), Stability AI (images), ElevenLabs (voice) and Google Sheets/Drive (content queue and assets) to generate each short automatically.

Impact & Value

  • 20+ shorts per month generated in production
  • 4 external APIs orchestrated in a single flow with no manual intervention
  • Content queue and assets managed from Google Sheets and Drive

Technologies

n8nOpenAI APIStability AIElevenLabsGoogle SheetsGoogle DrivePrompt Engineering
Public

Legacy-to-Modern Architect

Feb 2026

Problem

Legacy monolithic applications with accumulated technical debt, slow deployment times, and high friction for incorporating new features.

Architectural solution

Progressive migration architecture from monolith to microservices using strangler-fig pattern. CI/CD pipeline with GitHub Actions, Docker containers, and deployment on AWS/Azure. Incremental refactoring with feature flags for zero-downtime migration.

Impact & Value

  • 70% reduction in deployment time
  • Decoupled architecture ready to scale horizontally
  • Technical documentation of replicable patterns

Technologies

Node.jsDockerGitHub ActionsAWSAzureMicroservicesReactTypeScript
See all my repositories on GitHub

04. contact

Let's work together

Do you have an ambitious project or a position where you need LLM agents, RAG or QA automation in production? I am available for conversations.

Send me a message

I reply within 24h. Tell me about the technical challenge or opportunity and we will see if there is a fit.

diegoalonso139@gmail.com
GitHub LinkedIn