>> SYSTEM READY
ENGINEERING INTELLIGENCE.
RESEARCHING MULTIMODALITY.
BUILDING AGENTS.
CURRENT_STATUS●
Undergraduate Researcher @ Vector Institute. Focus on Foundation Models & Medical AI.
SHUOLIN(LEO) YIN
BASED IN TORONTO, CANADA
EST. 2023
InitializeEST. 2023
EXP.
PROFESSIONAL HISTORY
& RESEARCH
- Lead technical strategy and delivery of a purpose-built AI commercial engine for medical devices across ingestion, retrieval, reasoning, and automated outreach.
- Design a multi-agent system with MCP-based tool routing, secure enterprise connectors, audit logs, telemetry, and failure isolation for resilient integrations.
- Develop a compliance-first RAG stack with approved-source retrieval, automatic citations, and guardrails aligned to medical device regulatory and marketing constraints.
- Build document understanding for complex medical layouts (tables, symbols, multi-column, multilingual labeling) and scale OCR throughput via cloud batching and pipeline parallelism.
- Implement retrieval optimizations (vector indexing, hybrid search, reranking, prompt-retrieval co-design) to improve factuality and traceability.
- Train and deploy a compact outreach SLM on proprietary high-conversion data for high-precision personalization with minimal inference cost.
- Establish the production ML platform: model CI/CD, offline-to-online eval parity, drift detection, A/B experimentation, and tight customer feedback loops.
- Own and manage the engineering roadmap, aligning model research, reliability, and product milestones across early customers and pilots.
Multi-Agent SystemsMCP ToolingRAGVector/Hybrid SearchMLOps/CI-CD
- Train and deploy multimodal foundation models with multitask learning on large-scale 2D/3D medical imaging datasets.
- Develop baseline models and comprehensive preprocessing/quality-control pipelines for MICCAI FLARE 2025 (augmentation, QC workflows, fine-tuning protocols).
- Architect an end-to-end vision-language framework for automated whole-slide report generation with multi-scale features, cross-attention, and contrastive alignment across 10,500+ cases.
- Scale DINOv3 self-supervised training on 300M images across 39 categories using distributed data-parallel training, gradient accumulation, and mixed precision on H100 clusters.
- Author an educational framework paper on medical VLM development (MICCAI educational challenge finalist) and support experiment tracking, profiling, and documentation for publications.
PyTorchMultimodal LLMsDINOv3Multi-GPU ClustersData Pipelines
- Investigated MLLM hallucination reduction through systematic literature review and experimental design toward MICCAI 2025 publication.
- Developed automated data pipeline and custom annotation tools, reducing data processing time by 300% and improving annotation accuracy by 40%.
- Delivered weekly technical presentations and detailed progress reports to drive collaboration across the research team.
- Conducted PyTorch experiments on state-of-the-art MLLM architectures, evaluating hallucination mitigation methods and surfacing improvement opportunities.
Medical MLLMsPythonPyTorchData Automation
- Founded and scaled a non-profit through a four-tier management system, growing to 50+ members and 100+ global non-profits with 500+ monthly active users.
- Architected and developed a full-stack volunteer management platform (React Native + React.js) with CI/CD and Agile delivery for 15+ major features and 4 revisions.
- Secured Google Cloud partnership with cloud credits and technical training for 20+ developers; designed scalable Firebase/Firestore data architecture for 100+ org profiles.
- Implemented AI-powered volunteer matching and analytics for churn prevention, and led a tech mentorship program enabling 30+ projects and 10+ speaking engagements.
TensorFlowscikit-learnFirebaseTensorFlow.js
PROJECTS
SELECTED BUILDS
AI AGENTS
FULLSTACK
RESEARCH
01
React NativeRAGComputer Vision
Echo – AI-Powered Sustainable Fashion Marketplace
Lead Developer // Jan 2025
Secured 2nd Place in Global Finals at BCG & Global Spark “Hack the Globe” as sole developer.
02
OpenAI Agents SDKMCPRAG
EZ-Career – Autonomous AI Job Application Agent
AI Engineer & Full Stack Developer // April 2024 – June 2024
Architected a multi-agent AI system using OpenAI Agents SDK with an Agent-as-Tool pattern for end-to-end job application automation.
03
React NativeLangChainAWS Lambda
YiXing – AI-Driven Personalized Travel Planner
Project Lead / Co-Founder // Jun 2023 – May 2024
Led development of an AI travel planner using fine-tuned GPT models with LangChain-powered RAG for personalized itineraries.
04
React NativeFastAPIFew-shot Learning
ReassurED – LLM-Powered Emergency Care Navigator
Project Lead // Sep 2024
Led a 24-hour hackathon build of a cross-platform emergency healthcare app using React Native (Expo) with real-time Firestore hospital data.
TECHNOLOGIES
LIVEPythonJavaScript/TypeScriptC/C++C#SwiftMATLABHTML/CSSSQLBashLaTeXPyTorchTensorFlowHugging FaceOpenCVYOLOLangChainscikit-learnGCPAWS (Lambda)HPC ClustersDockerFirebaseGitCI/CDMulti-GPU ClustersVertex AIWeights & BiasesJupyterMLflowReact/React NativeNode.jsUnityFusion 360BlenderAutoCADPythonJavaScript/TypeScriptC/C++C#SwiftMATLABHTML/CSSSQLBashLaTeXPyTorchTensorFlowHugging FaceOpenCVYOLOLangChainscikit-learnGCPAWS (Lambda)HPC ClustersDockerFirebaseGitCI/CDMulti-GPU ClustersVertex AIWeights & BiasesJupyterMLflowReact/React NativeNode.jsUnityFusion 360BlenderAutoCAD
AutoCADBlenderFusion 360UnityNode.jsReact/React NativeMLflowJupyterWeights & BiasesVertex AIMulti-GPU ClustersCI/CDGitFirebaseDockerHPC ClustersAWS (Lambda)GCPscikit-learnLangChainYOLOOpenCVHugging FaceTensorFlowPyTorchLaTeXBashSQLHTML/CSSMATLABSwiftC#C/C++JavaScript/TypeScriptPythonPythonJavaScript/TypeScriptC/C++C#SwiftMATLABHTML/CSSSQLBashLaTeXPyTorchTensorFlowHugging FaceOpenCVYOLOLangChainscikit-learnGCPAWS (Lambda)HPC ClustersDockerFirebaseGitCI/CDMulti-GPU ClustersVertex AIWeights & BiasesJupyterMLflowReact/React NativeNode.jsUnityFusion 360BlenderAutoCAD