Real-Time Ingestion Pipelines
Continuous ETL pipelines ingesting public and spatial datasets into PostgreSQL vector and relational stores with deterministic verification.
Open to Principal / Lead Engineering Roles
I Build Production AI Systems End to End. From Infrastructure to Interactive Products.
I’m Jace K., a hands-on systems architect with 20+ years of professional engineering experience. I independently built TradeLike Land, a live bilingual platform combining continuously updated public data, deterministic analytics, knowledge-graph RAG, role-based multi-LLM orchestration, and spatial visualization.
20+ YearsEnterprise Systems & Architecture
1,000+ UsersOrganic Adoption Without Paid Marketing
End to EndInfrastructure, Data, AI & Frontend
Live PlatformProduction Korean & English SPA
Systems Architecture
Deterministic Data Systems Meet Generative AI.
Designed and engineered from bare-metal hardware to front-end visualization, prioritizing verifiable answers, source attribution, and deterministic analytics.
Knowledge-Graph & Hybrid Retrieval
Combining knowledge-graph retrieval with PostgreSQL vector and full-text search plus deterministic database tool execution for source-aware analysis.
Operational Retrieval LayerMulti-Tier LLM Orchestration
Task-routed models for classification, extraction, retrieval, translation, tool selection, and large-model reasoning across self-hosted GPU infrastructure.
Spatial & Map Interface
Low-latency, multilingual (Korean/English) Single Page Application delivering real-time map visualizations, scenario forecasting, and interactive analytics.
Live Operational System
TradeLike Land
A live bilingual real estate intelligence platform operating on continuously updated South Korean market data. It demonstrates end-to-end architecture across ingestion, hybrid retrieval, deterministic analysis, multi-model inference, and interactive spatial visualization.
- Real-time public data pipeline automation
- Knowledge-graph, vector, and full-text hybrid retrieval
- Deterministic tools for time-series comparison and forecasting
- Interactive map, chart engine, and source-attributed AI responses
- Multilingual production implementation (English / Korean)
ENGINEERING BACKGROUND
Built End to End by a Principal-Level Engineer.
TRADELIKE was engineered independently from the ground up—from bare-metal GPU server assembly and networking to backend AI pipelines, databases, and frontend spatial rendering. Before TRADELIKE, I worked across low-latency trading systems, bioinformatics pipelines, global B2B and e-commerce platforms, and high-concurrency enterprise services.
I am currently open to full-time Principal and Lead engineering roles where I can bring this end-to-end experience into a larger product and engineering organization.