Hanoi, Vietnam | Solution Architect • Data Platforms & AI Systems • Software Engineer
Software Engineer and Solution Architect working on enterprise data platforms, AI systems, and distributed infrastructure.
-
Data Platforms & Lakehouses: Building storage/compute-separated platforms with Apache Iceberg, MinIO/S3, and Apache Polaris. Implementing Medallion Pipelines (Bronze
$\to$ Silver$\to$ Gold) with data validation gates and end-to-end lineage via OpenMetadata. - AI Systems & Agents: Building LLM gateways, token optimization tools, multi-agent workflows, and RAG pipelines. Creator of dshmp.com (DeepSeek Harness Plugin Marketplace).
- Backend & Systems: Writing services and concurrent systems in Rust, Go, Python, and TypeScript.
- Security & IAM: Access control policies (OPA / ABAC), single sign-on (Keycloak SSO/MFA), and OAuth2 authentication.
-
dshmp.com — DeepSeek Harness Plugin Marketplace:
- Architected and built the verified discovery and catalog platform for the DeepSeek Harness (
dsh) plugin ecosystem. - Implemented strict manifest verification gates parsing
dsh.bundledirectly from the npm registry packument to eliminate noise and topic squatting. - Designed an automated ingestion pipeline with Python & Dagster for plugin discovery, mirror metadata enrichment, and star aggregation.
- Built a high-performance SSR web application with SvelteKit 2, Svelte 5, Tailwind CSS 4, and PostgreSQL 17 FTS (
tsvector+ trigram fuzzy search) with full multi-locale routing (en,vi,zh).
- Architected and built the verified discovery and catalog platform for the DeepSeek Harness (
-
City-Scale Shared Lakehouse Platform:
- Designed a storage/compute-separated Lakehouse using Apache Iceberg on MinIO / S3.
- Structured data with Medallion Architecture: raw ingestion (Bronze)
$\to$ standardized entities (Silver)$\to$ aggregated analytical views (Gold). - Built multi-source ingestion (APIs and form validation) with idempotent loads, automated quarantine rules, and metadata cataloging via OpenMetadata and Apache Polaris.
- Configured organizational hierarchy-based authorization with row-level filtering.
-
Regional Multi-Agency Data Platform:
- Unified data integration across multiple departments and administrative systems.
- Standardized Git-managed API contracts with automated spec validation and OAuth2 authentication.
-
Sectoral Data Network & Exchange Architecture:
- Built a hub-and-spoke exchange model with edge gateways connecting distributed data endpoints.
- Configured policy-based access control (OPA / ABAC), centralized IAM (Keycloak), and archival workflows.
Data Lakehouse & Streaming
AI & Machine Learning
Backend & Systems
Governance & Security
-18181b?style=flat-square&logo=openpolicyagent&logoColor=white){kind=icon}
Databases, Cloud & Infrastructure
