AboutExperienceEducationProjectsResearchSkills

Hi, I'm

Juan Bernal

Software & Industrial Engineer | AI Enthusiast

Architecting mission-critical systems and high-performance AI ecosystems through Systems Thinking and Operations Research.

About Me

I am a Software and Industrial Engineer specializing in the architecture of high-scale distributed systems and AI-driven ecosystems. By integrating Industrial Engineering principles, I treat software development as a rigorous optimization process where efficiency, scalability, and long-term reliability are engineered into the foundation.

I specialize in building high-performance, mission-critical software where excellence is the baseline. My approach combines the versatility of a full-stack architect with the precision of a research-driven engineer, ensuring that every system I design exceeds strict functional requirements while maintaining world-class standards for security and reliability.

Professional Experience

AI Engineer / Backend Engineer

Nebula Medical

August 2025 – February 2026

AI Engineer / Backend Engineer for a medical intelligence platform, developing AI models with leading LLMs, building RAG systems, and creating data pipelines for healthcare workflows.

  • Led backend architecture of an AI-driven medical platform using LangChain for multi-model orchestration across Claude, GPT-5, and Gemini, reducing average AI response latency by 35% through intelligent model routing and caching.
  • Designed and implemented a RAG-based conversational memory system with vector search (Pinecone), enabling persistent clinical chat context and reducing AI hallucination rate and improving diagnostic response accuracy.
  • Engineered a scalable Data Access Layer (DAL) with Prisma 7 and PostgreSQL, optimizing complex queries through advanced indexing and caching strategies, sustaining p99 < 500ms and p95 < 200ms under concurrent clinical workloads.
  • Implemented end-to-end encryption and HIPAA-compliant security protocols across all patient data flows, achieving no critical findings in our internal audits.
  • Integrated 3+ healthcare interoperability APIs (FHIR R4, SaludTools) enabling seamless data exchange across provider systems and reducing integration overhead by 50%.
  • Architected and deployed a multi-tenant authentication system using Clerk, reducing onboarding friction for new healthcare organizations by 70%.
  • Implemented API Gateway using AWS API Gateway and Lambda functions, reducing API response latency by 80% and reducing application bundle size.
Next.jsLangChainRAGPrismaPostgreSQLHIPAA

Intern - Software Developer I

Caseware

June 2024 – July 2025

Contributed to a global audit platform serving 475,000+ users across 130 countries, developing integrations in the imports/bindings module, backend microservices, and modern frontend architectures while leading technical migrations and agile ceremonies.

  • Engineered integrations within the imports/bindings module of the Caseware platform, a mission-critical data ingestion layer used by 475,000+ professionals across 130 countries, ensuring reliable, schema-compliant data mapping at enterprise scale.
  • Developed critical backend microservices in Java Spring Boot and Node.js and deployed serverless infrastructure on AWS through CDK, optimizing operational costs. Launched CDK projects from development to production environment.
  • Led massive data migrations (500k+ records) ensuring transactional atomicity and reliability. Improved data access patterns latency time by 40%
  • Implemented reactive components in Angular, reducing application load time by 30% through lazy loading and bundle optimization techniques.
  • Configured orchestrated deployments in Kubernetes
  • Responded to on-call production incidents with rapid triage and resolution for clients.
  • Led Scrum ceremonies for a 7-developer team, ensuring continuous value delivery for the imports team.
  • Worked with AWS services such as S3, DynamoDB, EC2, EKS, RDS.
JavaSpring BootAWS CDKAngularKubernetesScrum

Education

B.S. Systems and Computing Engineering

Universidad de los Andes

Bogotá, Colombia

Jan 2021 – June 2025

GPA 4.21 / 5.0

Relevant Courses

Transactional SystemSoftware architectureData structures and algorithmsBusiness IntelligenceAPI developmentObject-oriented programmingWeb development

B.S. Industrial Engineering

Universidad de los Andes

Bogotá, Colombia

Jan 2022 – June 2025

GPA 4.21 / 5.0

Relevant Courses

Advanced OptimizationMetaheuristicsProbabilistic modelsDiscrete simulationProbability and Statistics IProbability and Statistics IIFinancial EngineeringLogistics

Projects

Vista previa de Trove: Local Open-Source Search Engine & MCP Server (Contributor)

Trove: Local Open-Source Search Engine & MCP Server (Contributor)

A high-performance local search engine that indexes content across GitHub, Notion, Slack, and local files into a unified vector database. Includes an MCP server implementation that optimizes AI agent workflows by providing semantic search capabilities over local data.

TypeScriptNode.jsSQLiteElectronReactMCPTransformers.jsOllama
View Project
Vista previa de Aura AI: Intelligent Talent Matching Platform with NLP, GNN & LLMs

Aura AI: Intelligent Talent Matching Platform with NLP, GNN & LLMs

An end-to-end ML platform that transforms talent acquisition by replacing keyword matching with semantic fit. It features multi-modal document understanding, a learned skill taxonomy via Knowledge Graphs (Neo4j), and contrastive candidate-job matching using Graph Neural Networks (GNN). Orchestrated through a production-grade MLOps pipeline with LLM-powered agents for explainable scoring and reasoning.

FastAPINext.jsPyTorch (GNN)Neo4jQdrant (Vector DB)PostgreSQLMongoDBRedis/CeleryLangChain
View Project
Vista previa de CV Generator

CV Generator

A modular Python library for generating professional CVs in DOCX and PDF format.

Pythondocxreportlab
View Project
Vista previa de Sports Betting Prediction Engine. ML Pipeline with Kelly Criterion

Sports Betting Prediction Engine. ML Pipeline with Kelly Criterion

ML pipeline that implements temporal features from historical match data, trains and calibrates ensemble models (Random Forest, XGBoost, LightGBM), and sizes bets using the Kelly Criterion to exploit pricing inefficiencies in sports betting markets.

Pythonscikit-learnXGBoostLightGBMpandasNumPymatplotlibseaborn
View Project
Vista previa de Financial Market Alert WebSocket System

Financial Market Alert WebSocket System

Real-time financial market alert system using WebSocket technology. Implements event-driven architecture with Redis for message brokering and PostgreSQL for data persistence. The system ensures low-latency data delivery for time-sensitive financial operations.

TypeScriptNestJSRedisPostgreSQLWebSocketDocker
View Project
BI

Biological Predictive Modeling

Developed predictive models for complex biological data using machine learning techniques. Applied advanced preprocessing (PCA), class balancing, and Deep Learning (TensorFlow/Keras) to achieve 98% accuracy in predicting complex patterns.

PythonScikit-learnPandasNumPyTensorFlowKeras
View Project
EV

EV Charging Station Multi-Objective Optimizer

Developed a high-performance optimization system in Java and C++ for Electric Vehicle charging infrastructure. The system leverages multi-objective algorithms to minimize energy costs while maximizing customer service levels, resulting in a production-ready scheduling model.

JavaC++Optimization AlgorithmsData Modeling
Private project
Vista previa de Enterprise Management Platform (Hogar de Abuelos)

Enterprise Management Platform (Hogar de Abuelos)

A comprehensive management platform for facility administration, streamlining operations, records management, and resource allocation to improve operational quality.

PythonDjangoSQLiteHTML/CSS
View Project
MI

Microservice orchestration using Spring Boot and Docker

Architected a scalable distributed system using Java Spring Cloud (Eureka, Config, Gateway) within an Nx monorepo to centralize dependency management. Implemented asynchronous messaging via RabbitMQ and resilience patterns with Resilience4j, integrating JWT security and Zipkin for distributed tracing.

Spring BootDockerAWS
Private project

Research

Biological Predictive Modeling (B.S. Thesis)

2025 - I

Engineered deep learning architectures for high-dimensional biological data. Leveraged PCA for dimensionality reduction and Keras/TensorFlow for model development, achieving 98% accuracy in complex pattern recognition.

Analytics Forum 2025

2025 - I

Showcased undergraduate thesis findings at the 2025 Analytics Forum, presenting advanced machine learning applications in bioinformatics to an audience of industry experts and researchers.

COPA Research Group Presentation

2025 - I

speaker at the Center for Optimization and Applied Probability (COPA) to discuss the integration of neural networks and probabilistic modeling in biological systems.

Skills

Data-focused expertise across machine learning, statistical modeling, LLM applications, data pipelines, and production-ready AI systems.

Data Science

Pythonpandas / NumPyscikit-learnStatistical ModelingFeature EngineeringData VisualizationExperiment Design

Machine Learning & AI

LLM DevelopmentRAG SystemsLangChainPrompt EngineeringPyTorchTensorFlow / KerasXGBoost / LightGBMGraph Neural Networks

Data Engineering

Data PipelinesETL / ELTSQLData ModelingApache AirflowRabbitMQEvent-Driven Data Flows

Databases

PostgreSQLMySQLMongoDBDynamoDBRedisPineconeNeo4jQdrant

MLOps & Cloud

AWSDockerKubernetesGitHub ActionsModel Serving APIsMonitoring & Observability

Backend & APIs

FastAPINode.js / NestJSJava / Spring BootREST APIsGraphQLMicroservices

Soft Skills & Tools

English (Fluent C1)Technical CommunicationScrumDesign ThinkingConfluence & Jira

© 2026 Juan Bernal