Constantine Chung
Senior AI Engineer
I build AI systems that think, verify, and act — then I make sure they actually work in production. I'm drawn to hard problems where intelligence meets engineering: systems that reason over messy data, search across millions of records, and make decisions you can trust.
Experience
Senior AI Engineer — LFX Digitals Ltd.
Building end-to-end AI solutions — from document intelligence platforms and AI-powered search engines to agentic systems for data analysis, research, and automated reconciliation.
- Created an unified platform for document validation, extraction and content checking from images and PDFs
- Built AI agents for data analysis, visualisation, and database retrieval through a chatbot interface
- Implemented an AI-powered search engine enabling natural language and image searches over millions of entries
- Built AI-powered research tool for in-depth company background and news research
- Created ensemble learning system for commodity cost prediction
- Designed automatic reconciliation with NER powered by LLM
- Developed optimisation algorithm for automatic replenishment minimising storage cost
Financial Engineer — Lab for AI-Powered Fintech Ltd.
Developed ML and deep learning models for financial applications. Fine-tuned LLMs and built production pipelines for chatbots, RAG systems, and sentiment analysis.
- Fine-tuned LLMs for financial applications
- Built end-to-end ML pipelines for production deployment
- Integrated LLMs with chatbot, RAG, and sentiment analysis applications
- Developed financial models using machine learning and deep learning
Officer — NAW INC. LTD
Built databases for retail operations, managed web presence, and performed sales and inventory analytics.
- Built database infrastructure for retail business
- Managed website and online store operations
- Performed sales and inventory analysis
Education
King's College London
MSc Global Finance Analytics
2019 — 2021
London School of Economics
Certificate, Data Analysis
2019
University of Sydney
BA Economics
2011 — 2014
Certifications
Featured Projects
Deep agent framework for document intelligence
Converts PDFs and Excel files into validated, schema-consistent JSON for downstream workflows. Supports 100+ page documents and large tables with multi-page parsing. Handles long structured outputs up to hundreds of rows.
- Verification-and-repair loop to reduce silent extraction errors
- Structured outputs with aggregation, normalisation, and mapping logic
- Adopted in production by an international real estate developer
Agentic validation system for contract review
Cross-checks contract terms against internal system records to detect inconsistencies and missing fields across multiple contract sections.
- User-defined validation rules with cross-field consistency checks
- Traceable citations to contract clauses and internal records
- Handles multilingual contracts with standardised English outputs
Workflow-integrated deep agent for visual art
A visual art inspiration chatbot that generates paintings, analyses images, creates tutorials, and searches references — all through a conversational interface with adaptive memory.
- Workflow-integrated agent with plan-first execution and progress tracking
- Adaptive user profiling that evolves art preferences across sessions
- Conversation compression for long-running sessions without context loss
Enterprise AI Search Engine
Multi-modal search with unified embeddings
A production search engine supporting text, image, and image-with-text queries through a unified embedding space. Designed for enterprise scale with sub-linear retrieval complexity across millions of entries.
- Multi-modal retrieval — text, image, and image-with-text in a single pipeline
- Unified embedding model mapping images and text into one vector space
- Scalable search algorithm with sub-linear complexity for large-scale corpora
Skills
Agentic AI
Plan-and-execute architectures, context engineering, tool orchestration, memory and skill systems, verification-and-repair loops, multi-agent workflows
LLM Engineering
Prompt engineering, structured outputs, multi-model orchestration, guardrails, evaluation pipelines, fine-tuning (LoRA/QLoRA/PEFT)
RAG & Retrieval
Vector and hybrid search, reranking, multi-modal embeddings, knowledge graphs
Machine Learning & Deep Learning
Supervised and unsupervised learning, clustering, ensemble methods, time series forecasting, feature engineering, model evaluation and selection
Statistical Modelling & Optimisation
Bayesian inference, probabilistic programming, calibration, stochastic simulation, optimisation with constraints, linear programming
Production & MLOps
End-to-end ML pipelines, containerised deployment, cloud infrastructure, CI/CD, databases, API development
What I'm Reading
Books
Papers
