Daniel PHAM
I build production ML pipelines, geospatial analytics platforms, and agentic workflows. With 10+ years of experience from academic research to deployment, I work across the full stack from modeling to cloud infrastructure.
About Me
10+ years from academic research to production AI systems
I am a Data Scientist and AI/ML Engineer with a Ph.D. in Computer Science, based in Metz (France) on the Luxembourg border. I hold a valid Luxembourg work permit (fast-track employer transfer, no visa sponsorship required).
Over the past 10+ years, I have taught Computer Science and built R&D and production systems in Vietnam, France, and Luxembourg across Machine Learning, Geospatial Data Science, Mathematical Optimization, and Agentic AI, handling everything from modeling to cloud deployment.
I have shipped 10+ production AI/ML systems processing multi-terabyte climate datasets, 60,000+ IoT and satellite sensors, and industrial logistics data, improving pipeline speed and throughput by 20 to 40%.
Education
Ph.D. in Computer Science
University of Lorraine, Metz,
France 
Defended Nov 2023, conferred Feb 2024. Focus: ML optimization, ensemble learning, and LLM fine-tuning in finance & healthcare.
Master in ICT
University of Lorraine, France
& USTH 
Specialization: Environmental Decision Support Systems.
Bachelor of Science in MIS
National Economics University, Hanoi, Vietnam 
Dual focus: Computer Science and Business (Finance/Management).
Technical Expertise
Core technologies & production tools
LLMs, RAG & Agentic AI
ML, Geospatial & Optimization
Full-Stack, Data & MLOps
Professional Journey
2012 – 2026 · Luxembourg, France & Vietnam
Senior Data Scientist / Full-Stack Data Scientist
Databourg-Komunidad, Luxembourg
- Real-time Weather Forecasting (RainVision): Led development of multi-source forecasting models processing telemetry across 60,000+ IoT sensors and geostationary satellite feeds worldwide. Boosted accuracy by 5% via Deep Learning modernization of legacy ML.
- Infrastructure & Pipeline Modernization: Re-architected data pipelines with Airflow, RabbitMQ, and Docker, automating workflows and boosting system speed and throughput by 20 to 40%.
- Full-Stack GIS Dashboards: Delivered commercial-grade interactive map services (Django/FastAPI/MongoDB/MapboxGL) with secure APIs, adopted by public disaster management agencies across the Philippines, Indonesia, and Brazil.
- Global Climate Risk Platform (KCS): Engineered statistical estimation frameworks analyzing multi-terabyte netCDF datasets for long-term physical climate risk adaptation.
- Agentic AI & Modular Workflows: Built a multi-agent skills toolkit (Claude Code, GPT) to speed up team research, standardize peril pipelines, and share domain knowledge across teams.
R&D Software & Optimization Engineer (Postdoc)
LCOMS / LGIPM - Université de Lorraine, Metz, France
- Tech-Lead for "DCASOLVER": Led technical architecture, engineering, and deployment of the commercial web platform (DCASOLVER Demo) integrating CPLEX and mathematical optimization engines.
- Optimization & Routing Integration: Co-implemented and integrated optimization routines and constraint-handling modules for large-scale Vehicle Routing Problems (VRP) and GIS-based logistics routing.
- Full-Stack Deployment: Built and deployed scalable RESTful services using Python, Django, FastAPI, MySQL, and Linux.
Machine Learning R&D Engineer (Ph.D. Candidate)
LCOMS / LGIPM - Université de Lorraine, Metz, France
- Ph.D. Thesis: "Techniques d'apprentissage automatique basées sur DCA avec des applications dans la finance et la santé" (Defended Nov 2023).
- Applied ML & Optimization: Developed optimization-based architectures across Kernel SVMs, optimization-based Weighted Bagging, and fine-tuned transformer models (BERT / RoBERTa) to address severely imbalanced datasets in financial anomaly detection and clinical risk prediction.
- Industry-Defense R&D: Executed 3 joint research projects for naval defense contractor NAVAL Group, applying Deep Learning, Computer Vision, SLAM, and Robotics for autonomous path optimization.
Computer Science University Lecturer & Researcher
National Economics University, Hanoi, Vietnam
- Teaching: Taught Fundamental Computer Science, Applied Informatics, and Management Information Systems at Bachelor level.
- Research: Conducted academic research in Artificial Intelligence, Machine Learning algorithms, and applied information systems.
GIS Data Engineer & Web Developer
VinTech Group (Vingroup), Hanoi, Vietnam
- AI Logistics System: Engineered an AI-driven logistics and route tracking web platform with Django, Bootstrap, and Google Maps APIs for Vinmart retail operations across 2,000 stores nationwide.
- Data Architecture: Built automated data processing pipelines, designed relational database schemas in MySQL, and managed deployment on AWS.
ML Engineer & Product Owner (AI Lab)
NAL Jsc., Hanoi, Vietnam
- Team Leadership: Led an 8-engineer AI team delivering commercial AI applications and PoCs for the Japanese enterprise market.
- Key Systems: Delivered an Operator Chatbot, Iris Recognition & Pupil Tracking, Face Lock attendance system, and Hate Speech / Sentiment Detection systems.
NLP Engineer
Chappiebot Inc., Ho Chi Minh City, Vietnam
- AI Vehicle Search Platform: Developed search and ranking algorithms for an automobile search and recommendation platform (Otonhanh.vn).
- Vietnamese NLP Modules: Engineered end-to-end NLP tasks in Vietnamese: Named Entity Recognition (NER), topic classification, intent parsing, sentiment analysis, and dialogue systems.
Machine Learning Research Intern
LITA - Université de Lorraine, Metz, France
- Autonomous Marine Surface Vessels: Researched long-term autonomy and environmental data acquisition on Unmanned Surface Vessels (USV). Built simulation environments in Python/Pygame and reinforcement learning control models.
Featured Projects
Production geospatial software, meteorological AI & agentic systems
Global Climate Risk Intelligence Platform
Models and maps physical climate risks across global asset portfolios for climate adaptation planning. Fits Generalized Extreme Value (GEV) distributions on historical climate observations to estimate multi-decade return period hazards across standard climate scenarios.
Real-Time Rainfall Monitoring & Nowcasting Platform
Tracks and forecasts short-term precipitation (nowcasting) in real time for flood monitoring and disaster response. Integrates streaming telemetry from 60,000+ distributed IoT sensors and Himawari geostationary satellite feeds, adopted by public disaster management agencies across the Philippines, Indonesia, and Brazil.
DCASOLVER: Industrial Logistics & Route Optimization Platform
Interactive optimization web platform built on mathematical optimization algorithms developed at Université de Lorraine. Solves large-scale Vehicle Routing Problems (VRP), dynamic fleet dispatching, and multi-depot supply chain routing under operational constraints.
Commercial, Research & Engineering Projects
Production commercial deployments, academic & industrial research, internal team toolkits, and AI prototypes
Audit RAG Financial Compliance
Automates second-level NAV verification by reconciling UCITS/AIFMD fund prospectuses against Luxembourg CSSF circulars (02/77, 24/856). Uses table-aware chunking, bitemporal filtering, LangGraph multi-hop citation resolution, Hybrid Search with Cross-Encoder Reranking, and deterministic Pydantic/Decimal checks.
Climate Agent Kit
Internal multi-agent toolkit built to speed up research and automate peril modeling workflows across the team using specialized sub-agents (Claude Code, custom skills).
Drone Scheduling & Flight Simulation
Research project optimizing flight paths and allocation of aerial drone fleets under energy constraints. Developed an interactive Python/Pygame flight simulation environment to train reinforcement learning controllers.
Autonomous Vessel Guidance
Research project on long-term autonomy and environmental data collection for Kingfisher Unmanned Surface Vessels (USV) using ML control algorithms and simulation at LITA, Université de Lorraine.
Iris Recognition & Tracking
Commercial mobile computer vision system tracking mobile user gaze and engagement via iris position and pupil size with deep learning and real-time edge inference.
Face Access Control System
Internal biometric system automating employee check-in and attendance tracking with few-shot deep metric learning and live camera streams.
AI Vehicle Search Platform
Automobile search platform running Vietnamese Named Entity Recognition (NER), intent detection, topic classification, and sentiment analysis at Otonhanh.vn.
Certifications
Generative AI with LLMs
AWS (2024)MLOps with Azure
LinkedIn (2024)MLOps Essentials
LinkedIn (2024)Intro to Large Language Models
LinkedIn (2024)Deep Learning Specialization
Coursera (2018)Data Science in Healthcare
Univ. of Edinburgh (2023)Data Science Ethics
Univ. of Michigan (2023)Publications
A Block Coordinate DCA Approach for Large-Scale Kernel SVM
View DOIDCA-Based Weighted Bagging: A New Ensemble Learning Approach
View DOICost-sensitive weighted bagging DCA based method for imbalanced financial data
NUMTA Conference PresentationTechniques d'apprentissage automatique basées sur DCA
View Official Thesis PDFContact
I am currently open to Data Scientist, Machine Learning / AI Engineer, and Data Engineer roles.