05/11/2026 updated

JP
100 % available

AI Architect & Software Engineer | LLM, RAG, Multi-Agent Systems

Angeles City, Philippines
Only remote
Bachelor of Science, Computer Science
Angeles City, Philippines
Only remote
Bachelor of Science, Computer Science

Profile attachments

JoshuaPolicarpio.pdf

About me

I’m an AI Architect and Engineer passionate about designing intelligent systems that perform at the edge of what’s technically possible. I’ve built production-grade LLM, RAG, and multi-agent systems that have reduced latency by over 80%, boosted operational efficiency by 40%.

ARM ArchitectureData ArchitectureDigital ArchitecturePython (Programming Language)Software ArchitectureAzure Machine LearningNext.jsModel-Driven ArchitectureReact.jsFlaskLarge Language ModelsMulti-Agent SystemsFastAPIArtificial Intelligence Markup Language (AIML)
AI & LLM Engineering
End-to-end development of AI systems leveraging Large Language Models (LLMs), RAG architectures, and multi-agent frameworks. Proven track record of deploying self-hosted LLM infrastructures using models such as Qwen3-Coder-480B-A35B-Instruct, LiteLLM, and Crush, achieving measurable reductions in wait times and operational efficiency gains.

Machine Learning & Deep Learning
Extensive experience with PyTorch, TensorFlow, Keras, and Hugging Face for building and fine-tuning models including BERT, Llama, Gemma, and Anthropic Claude. Application of NLP, computer vision, and ensemble methods to solve real-world business and healthcare challenges.

Cloud Infrastructure & MLOps
Hands-on expertise with AWS (SageMaker, DynamoDB, RDS), Google Cloud Platform (BigQuery, Cloud SQL), and Microsoft Azure. Proficient in Kubernetes, Docker, MLflow, Kubeflow, Databricks, and Apache Spark for scalable, production-grade AI deployments.

Python & Backend Development
Strong proficiency in Python, Golang, and Rust for building high-performance backend systems and data pipelines. Experience with SQL, Pandas, NumPy, and Scikit-learn for data processing and analysis.

Prompt Engineering & Fine-Tuning
Expertise in prompt engineering, model fine-tuning, and building continuous learning pipelines with curated datasets to improve model accuracy and relevance in production environments.

Data Visualization & NLP
Proficiency in data visualization tools and NLP techniques including OpenCV, Gensim, and LangChain. Experience developing analytics platforms that enable non-technical users to generate SQL queries and data visuals using plain English.

HIPAA Compliance & Data Security
Demonstrated ability to build HIPAA-compliant data infrastructures with granular access controls, full audit trails, and secure multi-tenancy systems using Keycloak, ensuring 100% data isolation and zero security breaches.

WebRTC & Real-Time Communication
Implementation of WebRTC solutions for real-time communication features, contributing to innovative application capabilities and enhanced user interaction.

Big Data & Analytics
Experience with Hadoop, Apache Spark, BigQuery, KNIME, and SAS for large-scale data processing, reporting, and analytics pipeline architecture.

Languages

EnglishFluent

Project history

Co-founder

CycleSyncAI Labs
Engineered a low-latency conversational voice agent achieving sub-300ms response time and reducing average customer wait times by 80%. Architected and deployed a secure, self-hosted LLM infrastructure enabling 50+ team members to utilize internal models with 100% data security compliance. Designed and deployed the core AI model for a Lead Magnet Generation SaaS platform, boosting conversion accuracy by 35%. Implemented centralized logging and behavioral tracking for all LLM interactions, identifying 15% cost-saving opportunities. Designed a dual-memory system for AI agents using Redis, increasing user satisfaction scores by 25%. Developed an automated customer support agent with a dynamic RAG system resolving 85% of unique customer inquiries without human intervention.

AI Engineer

Shypv
Developed Logistics Chatess, an AI analytics platform enabling non-technical users to generate SQL queries and data visuals using plain English. Architected a secure multi-tenancy system with Keycloak for 50+ clients ensuring 100% data isolation and zero security breaches. Engineered a SQL Agent handling 85% of data requests autonomously, reducing analytics team workload by 60%. Improved SQL Agent accuracy by 40% via a continuous learning pipeline. Implemented a hybrid search system resolving over 90% of ambiguous user questions. Pioneered Medical Chatess, an AI diagnostic assistant integrated with EMRs, cutting initial patient review time by 30% for clinicians. Built a HIPAA-compliant data infrastructure for 10+ clinics and developed an AI analysis engine achieving a 92% clinical relevance score.

Software Engineer

Exilion Technologies Inc
Executed a character-based transformer spelling correction model with Triton and ONNX, reducing inference latencies to below 178ms at 35 requests per second. Conceptualized an AI-driven healthcare chatbot using LLM and Reinforcement Learning, resulting in a 25% increase in patient engagement and 30% reduction in customer service response times. Ensured HIPAA compliance, reducing data breaches by 28.5%. Collaborated with data engineers and scientists to architect scalable data pipelines. Implemented WebRTC solutions for real-time communication.

Software Engineer

Dreamers Inc
Developed and implemented advanced NLP algorithms and machine learning models, enhancing departmental decision-making by 14.1%. Specialized in delivering MVPs for cross-platform mobile applications with secure authentication and messaging features. Collaborated with cross-functional teams to integrate APIs using Firebase and PostgreSQL. Leveraged LLMs like OpenAI GPT for text generation, sentiment analysis, and language understanding, increasing processing speed by 19.2%. Engaged in continuous learning to stay updated on AI advancements, enhancing capabilities in data science and NLP by 9.4%.

Intern & Software Engineer

Academic Labs
Calculated document similarities using Gensim indexing and queries, improving search speed by 24.8%. Architected data-driven solutions with machine learning techniques, reducing overhead costs by 34.1%. Optimized predictive model efficiency with ensemble methods, improving accuracy by 8.9%. Increased reporting system accuracy by 4.8% through improved data validation. Delivered configuration management tools saving 27.7% of initial machine setup time. Developed monitor reports using in-memory cache, improving real-time data access by 10.6%. Integrated SaaS invoice solutions and payment functionalities via effective API utilization.

Contact form

Log in to get in touch

You need to be logged in to use the contact form.

Sign upLog in