Full-stack Developer · Edge AI & Distributed Systems

Hi, I'm Vo Thanh Tung

Architecting Scalable Microservices, High-Performance Web Applications & Edge AI Systems

Full-stack developer with 4+ years of production experience building scalable, enterprise-grade web applications and distributed systems using NodeJS, ReactJS, and TypeScript. Passionate about client-side Edge AI (WASM speech recognition & neural TTS), real-time streaming, and resilient cloud architecture that reduces operating overhead while delivering 60 FPS user experiences.

4+
Years Experience

Production web & backend engineering across streaming, telematics, and e-commerce.

8+
Microservices Built

Engineered scalable distributed services with SSO, RabbitMQ, and Elasticsearch.

~60%
Operating Cost Cut

Custom auto-scaling video pipeline slashing cloud video processing bills.

120k+
In-Memory Words

Full-text phonetic corpus indexed with sub-millisecond client-side fuzzy search.

$0
Cloud Speech Cost

Edge AI architecture running 100% in-browser Kaldi WASM & Piper Neural TTS.

Production Engineering Showcase

Personal Projects & Edge AI Breakthroughs

Architected to eliminate cloud latency and recurring API bills. Explore authentic production UI screenshots, client-side WebAssembly runtimes, and interactive feature walkthroughs.

Edge AI & WebAssembly$0 Cloud Cost100% Client Offline PWA

LingoCore🎙️

Privacy-First, Zero-Latency Offline English Pronunciation & Phonetics Coach powered by Edge AI

Traditional pronunciation platforms rely on cloud speech APIs, introducing latency (>300ms), privacy concerns, and high recurring bills. LingoCore eliminates all three by executing Kaldi speech recognition (Vosk WASM), Neural TTS (Piper WASM + ONNX), and in-memory phonetic search directly in the browser with zero server costs.

ReactTypeScriptViteTailwind CSSWebAssembly (WASM)vosk-browserPiper Neural TTSonnxruntime-webOPFSOrama SearchHowler.jsPWA Workbox
Operating Cost$0 (Client Edge)
Corpus Size120,000+ Words
Speech Latency<15ms (Zero RTT)
PWA Readiness100% Offline

Core Engineering Highlights

Client-side Kaldi WASM running vosk-model-small-en-us with constrained phrase grammar (setGrammar) for hallucination-free scoring.
Dual-tier speech synthesis: Piper ONNX neural voice models cached in Origin Private File System (OPFS) with instant Web Speech API fallback.
120k+ word phonetic dictionary combining CMU Pronouncing Dictionary and Brown Corpus frequency rankings.
Web Worker search parsing + TanStack Virtual for sustained 60 FPS scrolling through thousands of query matches.
Visit
LingoCore landing and edge architecture overview
Click to expand high-res screenshot
Landing & Edge Architecture Overview
Client-side Kaldi WASM + Piper Neural TTS ($0 Cloud Cost)

Privacy-first pronunciation coach executing Kaldi speech recognition & Piper Neural TTS 100% offline in browser.

Screenshots (1 of 5) — Click thumbnail to preview:
Career Milestones

Production Engineering & Enterprise Systems

Delivering high-scale microservices, automated video transcoding pipelines, geospatial telematics platforms, and enterprise workflow systems with measurable business impact.

Paracel Technology Solutions

Full-stack Developer

2023 – PresentDa Nang, Viet Nam

Core engineer developing high-scale microservices, telematics platforms, and legal workflow automation systems, serving hundreds of thousands of users and processing enterprise workloads.

Key Production Systems Engineered
~60% reduction in video processing cloud spend
300+ videos/day processed automatically
Sub-second search across 100k+ records
Key Responsibilities & Architecture:
  • Engineered end-to-end features across 8 modular services, from REST/event API contracts to modern Next.js frontend interfaces.
  • Implemented a centralized Single Sign-On (SSO) authentication service to unify user identity, tokens, and RBAC across all ecosystem services.
  • Architected an auto-scaling video transcoding pipeline processing 300+ videos/day, reducing per-video cloud transcoding costs by ~60% vs AWS MediaConvert.
  • Integrated RabbitMQ for async distributed worker jobs and Elasticsearch for full-text search, achieving sub-second discovery across 100k+ video records.
  • Built multiple responsive frontends: primary streaming platform, video transcoding ops dashboard, live shopping checkout UI, and embeddable real-time comment widgets.
Technology Stack:
TypeScriptExpressJSNextJSPostgreSQLMongoDBRedisElasticsearchRabbitMQAWS MediaConvertS3 / MinIO / Backblaze B2
Sub-50ms geospatial query response times
100% decoupled async push notification fanout
Key Responsibilities & Architecture:
  • Designed and implemented geospatial reporting system using bounding-box queries for high-performance spatial data retrieval over millions of GPS telemetry points.
  • Engineered asynchronous push notification infrastructure using BullMQ queues and FCM topic broadcasts, decoupling notification fanout from HTTP request cycles.
  • Built interactive map-based frontend with automated viewport bounds fitting and custom SVG markers for smooth trip and driving event visualization.
  • Developed dynamic analytics dashboards with configurable filters, risk score calculation engines, and interactive trend charts.
Technology Stack:
TypeScriptNestJSNextJSPostgreSQLRedisBullMQFirebase Cloud Messaging (FCM)Twilio SMSGoogle Maps ServicesDigitalOcean Spaces
Automated multi-step document intake workflows
Generic reusable data table utilized across 15+ internal screens
Key Responsibilities & Architecture:
  • Engineered automated email ingestion service retrieving and parsing emails and attachments with background task pipelines for case assignment and document processing.
  • Developed a dynamic, modular form builder enabling legal operators to configure customizable workflows and client intake forms without redeploying code.
  • Created a generic, high-performance data table component with resizable columns, keyboard navigation, and cross-component drag-and-drop capabilities.
Technology Stack:
TypeScriptNestJSNextJSPostgreSQLSynology Network StorageRESTful APIs

Rikkeisoft

Full-stack Developer

2022 – 2023Da Nang, Viet Nam

Engineered full-stack solutions for enterprise clients in document management and high-traffic e-commerce domains.

Key Production Systems Engineered
Infinite scroll masonry layout for large media collections
Enterprise branding & media management dashboards
Usage analytics & customer activity reporting
Key Responsibilities & Architecture:
  • Implemented an infinite scroll masonry (brick) layout for efficient rendering of large media collections.
  • Built admin dashboard features for customer-specific branding, including custom banners and logos.
  • Developed media management interfaces for uploading, organizing, and maintaining large image libraries.
  • Created analytics and reporting dashboards to track media usage and customer activity.
  • Integrated RESTful APIs to connect frontend features with backend services.
Technology Stack:
ReactJSNextJSPHPMySQLRESTful APIs
Automated OCR extraction pipeline with regex parsing
Dynamic editable templates reducing manual data entry
Multi-step document upload, validation & editing workflow
Key Responsibilities & Architecture:
  • Implemented data extraction logic using regular expressions to parse and structure OCR output from Google OCR.
  • Built dynamic document templates to auto-populate extracted data, reducing manual data entry effort.
  • Designed and developed a multi-step document processing workflow for file upload, extraction, validation, and editing.
Technology Stack:
TypeScriptNodeJSNestJSReactJSMongoDBGoogle OCRRESTful APIs
Technical Capabilities

Skills & Engineering Matrix

Full-spectrum capabilities spanning high-performance client runtimes (WASM, React 19, Monaco), robust backend architectures (NestJS, RabbitMQ, BullMQ), and distributed cloud infrastructure.

Frontend Engineering

Next.js

SSR, SSG, Server Actions, Dynamic OG

React

Hooks, Virtualization, Concurrent Mode

TypeScript

Strict type systems, Generics, ASTs

Tailwind CSS

@theme tokens, modern responsive UI

Web Workers & PWA

Off-thread compute, Service Workers

Backend & Microservices

NestJS & Express.js

Modular architecture, REST & GraphQL

NodeJS Runtime

Asynchronous I/O, Event loop, Stream pipes

Microservices Architecture

Service mesh, distributed decoupling

SSO & Security

OAuth 2.0, JWT, RBAC, Data sanitization

Message Queues

RabbitMQ job workers, BullMQ async queues

Edge AI, Audio & Media

WebAssembly (WASM)

Browser-level high-performance runtimes

Vosk Browser (Kaldi)

Offline speech recognition & grammar

Piper Neural TTS & ONNX

In-browser neural speech synthesis

Origin Private File System

OPFS local neural model caching

Video Transcoding Pipelines

Auto-scaling cost-saving transcoding

Databases & Search

PostgreSQL

Relational modeling, Indexing, Geospatial queries

Redis

Caching, BullMQ persistence, Pub/Sub

MongoDB

Document modeling, Aggregation pipelines

Elasticsearch

Full-text indexing, sub-second queries

Orama

In-memory search

Cloud, DevOps & Storage

Docker & Containerization

Multi-stage builds, dev/prod parity

AWS Cloud (S3, MediaConvert)

Scalable compute, storage pipelines

S3-Compatible Storage

MinIO, Backblaze B2, DigitalOcean Spaces

Git & CI/CD

Git flow, automated test pipelines

Firebase (FCM)

Topic broadcast & user push notifications

Get in Touch

Let's Build Something Extraordinary

Whether you are looking to scale an enterprise microservices ecosystem, implement zero-cost Edge AI speech pipelines, or build responsive web applications, I am open to discussing new opportunities.

Send a Direct Message
Da Nang, Viet Nam
UTC+7
Local Time:Loading...