ABOUT ME
Published AI researcher (Best Paper Award, ASIP 2025) and production AI engineer with a track record of shipping systems that serve real users at scale. Architected semantic search for 6,000+ Cambodian teachers (132% accuracy improvement) and built a multi-tenant RAG chatbot SaaS with 264/264 tests passing — deployed end-to-end without a separate ops team — leveraging AI coding agents (Claude Code, Gemini CLI) to ship faster without sacrificing quality. Specialize in LLM integration, RAG pipelines, and full-stack AI deployment. 5+ years of DevOps across Docker, Nginx, AWS, and multi-cloud environments. Pursuing a Master's in AI at CamTech University researching CNN-Transformer hybrid architectures for real-time anomaly detection. Multilingual: Khmer, English, Chinese.
PERSONAL DETAILS
Nationality: Cambodian
LANGUAGES
Khmer
English
Chinese
LINKS
HOBBIES
AI Research, Programming, Movies, Reading
REFERENCE
Dr. May Thu
Dean of Faculty of Engineering, Cambodia University of Technology and Science (CamTech)
Contact available on request
Monireach Tang
AI Engineer
EDUCATION
Cambodia University of Technology and Science (CamTech)
Master's Degree, Artificial Intelligence
Oct 2024 - Dec 2027
- Research Topic: "Contactless AI-Powered Abnormal Movement Detection for Elderly Safety Assurance"
- Developing non-invasive AI system leveraging Computer Vision, Edge AI, and privacy-by-design principles for real-time anomaly detection
- Processing and annotating sensor data, performing feature engineering, and optimizing hybrid model (CNN-LSTM-Transformer)
- Best Paper Award, ASIP 2025 Conference — published in Asian Journal of Innovation and Policy (Privacy Governance in AI-Powered Elderly Monitoring)
- Key Technologies: Python, PyTorch, FastAPI, Computer Vision, Anomaly Detection, Edge Computing
Royal University of Fine Arts (RUFA)
Bachelor of Architecture (BArch)
Oct 2013 – Oct 2018
AI PROJECTS
Smart Customer Service Chatbot
LLM-Powered Chatbot Microservice
Live | monireach.com | HuggingFace Spaces
- Leveraged AI coding agents (Claude Code, Gemini CLI) to accelerate development velocity while maintaining production standards — TDD, CI/CD, and full test suite coverage throughout
- Built production-grade multi-tenant SaaS: API key auth, tenant-scoped DB isolation, and per-tenant system prompt (persona, tone, rules) — any business can configure and embed their own chatbot instance without writing AI code
- Implemented full RAG pipeline: document ingestion, OpenAI embeddings, ChromaDB vector store, semantic retrieval, and prompt injection — domain-specific responses per client without model fine-tuning
- Engineered hierarchical context memory: verbatim window for recent turns + async background summarization when token budget is exhausted — long conversations stay coherent at scale
- Integrated Groq LLM (Qwen3-32b) with SSE streaming pipeline for real-time typewriter UX;
implemented
<think>block parsing to cleanly separate reasoning traces from user-facing responses - Shipped
<smart-chatbot>web component (Preact + Shadow DOM) with CSS theming; deployed to HuggingFace Spaces via GitHub Actions CI/CD — 78/78 tests passing (pytest TDD) - Retained hybrid NLP routing layer (TF-IDF + Naive Bayes intent classification) before LLM invocation — combining classical ML precision with LLM generative capability for production reliability
Fraud Detection System
Production ML System with REST API
Production-Ready | GitHub
- Built end-to-end fraud detection system handling full ML lifecycle from data preprocessing to production API deployment
- Achieved 96% precision and 91% F1 score through custom threshold optimization and business cost-aware metrics
- Designed production-grade REST API with FastAPI featuring health checks, error handling, SHAP-based model explanations, and real-time predictions
- Containerized with Docker and orchestrated deployment pipeline with environment-specific configurations (dev/staging/production)
WORK EXPERIENCE
Manulife Cambodia
Generative AI Specialist
Jun 2026 - Present | Phnom Penh, Cambodia
- Design and implement GenAI solutions leveraging LLMs, multi-agent systems (LangGraph, Pydantic), and advanced orchestration frameworks for enterprise-scale deployment
- Build and fine-tune Generative AI models for text generation, summarization, and automation; implement Retrieval-Augmented Generation (RAG) for enterprise knowledge integration
- Identify and prioritize GenAI use cases aligned with business objectives; collaborate with stakeholders to translate requirements into scalable AI solutions
- Ensure adherence to data privacy and ethical AI standards; implement robust testing strategies for prompt evaluation, schema validation, and workflow correctness
- Partner with Data Engineers, MLOps, and QA teams for production deployment and monitoring; provide mentorship on GenAI best practices
Proserve Performance Co., Ltd.
Chatbot Specialist
Jun 2025 - Feb 2026
- Design and deploy automated chat solutions using NLP tools, creating conversation scripts and FAQs in Khmer language
- Monitor chatbot performance metrics, analyze user interactions, and optimize chat functionality for customer satisfaction
- Provide Mandarin language support for chat automation and implement workflow process improvements
Taishan Technology Co., Ltd.
DevOps & Project Coordinator
Dec 2023 - Jun 2024
- Managed deployment infrastructure including MySQL, PHP, Node.js, and NGINX servers, handling domains, DNS, SSL certificates, and system monitoring via Tencent Cloud tools
- Coordinated project requirements through Zentao and implemented Docker with NGINX proxy for streamlined multi-project deployment and management
Maya Technologies Co., Ltd.
Flutter Dev & DevOps
Dec 2020 - Dec 2023
- Collaborated with Java backend developers for seamless integration and configured servers across multiple cloud platforms
- Built Android APKs and iOS PWAs, managing versioning and deployment via Git/GitLab to deliver scheduled updates with improved workflows
VOLUNTEER EXPERIENCE
R4C (Tuktuk Charity)
Volunteer AI Engineer (Ad-hoc)
Aug 2025 - Present | Production | Staging (with AI Chatbot)
- Architected AI-powered educational platform serving 6,000+ Cambodian teachers, improving resource discovery accuracy from 25% to 58% (132% improvement) through hybrid search combining semantic embeddings and keyword matching with <2-second response time
- Designed multi-cloud AI architecture integrating Google Cloud services: OpenAI for semantic search, Gemini for chat and translation, Google Speech-to-Text for voice input, and Pinecone for 1,400+ educational resources
- Built microservices ecosystem using FastAPI and Docker: Khmer tokenization service (khmernltk) for NLP word segmentation, voice transcription service (FFmpeg + Google Cloud STT) for audio-to-text conversion, enabling voice-powered search interface
- Implemented bilingual embeddings system (Khmer-English) with intelligent caching and cost optimization (Gemini 2.5 Flash-Lite), reducing translation costs by 90% while maintaining search accuracy for both languages
- Developed RAG-based conversational chatbot with custom context filtering, dynamic URL mapping, and environment-aware responses, deployed on WordPress with AI Engine Pro integration