I build AI systems combining multimodal models, autonomous agents, computer vision, and cloud infrastructure.

My work focuses on implementing tool-using agents, evaluating model behavior, building reproducible experiments, and developing practical AI applications.

Specialties: Agentic AI · Multimodal Systems · Computer Vision · RAG · MCP · LLM Evaluation · Python · AWS

Open to opportunities in: AI Engineering, Applied Machine Learning, Multimodal AI, and Agent Systems.

LinkedIn

Meta Global AI Hackathon 2026

Research investigating whether general-purpose multimodal AI models can autonomously play unfamiliar 3D games and produce verifiable evidence for game quality assurance.

• Built an autonomous ViZDoom gameplay agent powered by Meta Muse Spark 1.3, with experimental SAM 3.1 visual grounding.
• Evaluated navigation, combat, and objective completion without reinforcement learning or game-specific fine-tuning.
• Developed gameplay telemetry, experimental sweeps, trace verification, controlled fault injection, and QA reporting.
• Demonstrated successful autonomous objective completion while documenting navigation failures and model limitations.

Python Multimodal AI Agent Systems ViZDoom AI Evaluation

Retrieval-augmented generation system built around a structured knowledge base of 202 yoga poses.

• Benchmarked BM25, dense embeddings, and hybrid retrieval against a 75-question evaluation set.
• Achieved 66% MRR with hybrid retrieval.
• Evaluated multiple language models for answer generation.
• Integrated Streamlit, PostgreSQL logging, Grafana monitoring, and Docker Compose.

Python RAG Retrieval Evaluation PostgreSQL Grafana Docker

Model Context Protocol server for reproducible computer vision augmentation workflows.

• Exposes image augmentation operations through structured JSON-RPC interfaces.
• Supports deterministic random seeds, strict validation, and optional vision-language verification.
• Designed for tool-driven image processing and repeatable experimentation.

Python MCP Computer Vision Albumentations JSON-RPC

Agent-oriented knowledge graph application for exploring and visualizing structured relationships.

• Built with FastAPI and interactive PyVis graph visualization.
• Explores dynamic knowledge representation and agent-driven information workflows.

Python Knowledge Graphs FastAPI PyVis Agent Systems

Serverless cloud infrastructure project implementing a personal website and visitor counter on AWS.

• Built using S3, CloudFront, Route 53, ACM, Lambda, API Gateway, and DynamoDB.
• Implemented GitHub Actions CI/CD with AWS OIDC authentication.

AWS Serverless Infrastructure CI/CD GitHub Actions

Research-oriented framework exploring multi-agent orchestration for edge computing and next-generation telecommunications.

• Investigates agent coordination and decision-making in edge environments.
• Explores integration of AI agents with mobile edge computing infrastructure.

Agent Systems Edge AI AWS 5G/6G

Development team leadership and contributions to an active Discord community automation project.

• Implemented event management and scheduling workflows.
• Developed continuous Pomodoro study sessions with custom patterns, participation tracking, and leaderboards.
• Improved community engagement tools, persistent state management, and bot reliability.

JavaScript Node.js Discord.js MongoDB Docker

Multi-agent simulation exploring Korean consumer behavior and cultural personas using CrewAI.

Python CrewAI Multi-Agent Systems

• 300 Days of Code 2025 — GenAI, RAG, MCP, LLMOps, cloud engineering, and agent systems.
• 200 Days of Code 2024 — Computer vision, deep learning, reinforcement learning, GANs, and graph neural networks.
• 100 Days of Cloud — AWS, cloud networking, containers, security, and deployment.

Education: MSc Mechanical Engineering, University of Exeter · MBA, Quantic School of Business and Technology

Certifications: AWS Certified Cloud Practitioner · AWS Certified AI Practitioner

📫 Contact: LinkedIn