AI Engineer and Agentic AI Architect with 7+ years of experience designing and shipping production-grade AI systems, multi-agent pipelines, and full-stack AI products. I specialize in building intelligent agents that go far beyond simple retrieval.

I work across the full AI stack — model training (NLP/CV), MLOps, Databricks data platforms, and cloud-native deployment (AWS, Azure, GCP).

Agentic AI / LLMs
Multi-Agent Orchestration Skill-Based Agent Design Agentic RAG Human-in-the-Loop LLM Tool Use / Function Calling ReAct / Plan-and-Execute LangChain Transformers BERT NLP Computer Vision

MLOps / Infrastructure
Databricks MLflow Docker Terraform CI/CD (GitHub Actions) ONNX PySpark Unity Catalog

Languages & Frameworks
Python TypeScript SQL Next.js React FastAPI PyTorch TensorFlow

Cloud & Data
AWS Azure GCP MongoDB Atlas Power BI DBT Tableau

Creator & Full-Stack Engineer · July 2026 – Present
Bun TypeScript Hono Next.js React PostgreSQL Drizzle ORM Redis Redis Streams Better Auth

Open-source three-layer LLM gateway (proxy · ingest worker · control panel) that sits between applications and upstream model APIs.

• OpenAI-compatible & Anthropic surfaces with hierarchical child API key auth
• Encrypted provider credentials + model-alias routing for safe multi-tenant access
• Redis Streams ingest worker with at-least-once delivery into PostgreSQL (latency, cost attribution)
• Next.js control panel for provider/model registry, key lifecycle, and workspace governance

Creator & Full-Stack AI Engineer · Jan 2026 – Present
Python AWS Batch LangChain Next.js PostgreSQL Redis Docker

Autonomous self-reflective multi-agent platform for US equity paper trading.

• Supervisor–subordinate CIO topology orchestrating 7 specialist agents that continuously self-reflect and evolve
• +46% return over 6 months (vs QQQ +25%) — Performance Dashboard
• Real-time multi-region Redis streaming + AWS Batch for scalable, cost-efficient agent workloads
• Custom LangChain middleware for full decision traceability and transparent AI reasoning feed

Creator & Full-Stack Engineer · Jun 2024 – Present

Self-referencing tree models supporting unlimited nested hierarchies. Developers can write, publish via API, and deepen technical understanding in one platform.

• Senior Data Scientist — AI Engineering & Analytics · Zurich Insurance
• Software Engineer — Search & NLP · GLG · Hong Kong
• Fraud Data Scientist · Citi · Global/HK
• Data Mining Engineer — InfoSec · Tencent · China

• SOTA Behaviour Sequence Transformer for Movie Recommendation
• Accelerated HuggingFace CLIP with ONNX Runtime (3.5× GPU speedup)
• Graph Network Embedding (DeepWalk) with Incremental Training
• Full MLOps Recommendation API → AWS Lambda (MLflow + Prefect + Terraform + GitHub Actions)

I'm always open to discussing Agentic AI, multi-agent systems, LLM infrastructure, or interesting engineering challenges.

LinkedIn · Email · GitHub

⭐️ From Nelson Lin