AI & Machine Learning Consultant, London

Enterprise AI development, LLM integration, and computer vision solutions. Delivering production-ready ML systems for UK businesses.

AI Analytics Dashboard showing real-time machine learning insights

Machine Learning Services

End-to-end AI consulting from strategy to production deployment

LLM Fine-tuning & RAG Systems

Custom large language model development and retrieval-augmented generation systems. Multi-agent architectures with grounded, low-latency responses. Enterprise chatbots, document processing, and knowledge management solutions.

Computer Vision Development

Image recognition and video analytics for manufacturing and retail. Defect detection systems achieving 98% precision at 0.2-second inference. Object detection, visual inspection, and quality control automation.

MLOps & Pipeline Development

Production ML infrastructure on Google Cloud Platform and AWS. Automated training pipelines with Vertex AI, Apache Airflow, and CI/CD integration. Model deployment, monitoring, and continuous retraining workflows.

Predictive Analytics & Recommendations

Data-driven decision making through predictive modelling and recommendation engines. Customer segmentation, demand forecasting, and personalisation systems with 92% precision driving measurable business growth.

AI Strategy & Consulting

Strategic AI roadmaps aligned with business objectives. Opportunity assessment, technical feasibility analysis, and implementation planning. ROI-focused consulting with clear milestones and deliverables.

Healthcare & Medical AI

Clinical AI applications with regulatory compliance. GCP-certified with research experience in Parkinson's disease treatment prediction. Medical data analysis, clinical decision support, and healthcare automation.

Case Studies & Projects

Production AI deployments across manufacturing, retail, and healthcare

Multi-Agent Customer Support System

Personal project | Open source

A full, working multi-agent support agent using Google Gemini — a Policy Agent (RAG) and a tool-using Ticket Agent — built end to end for a fictional company to show the architecture.

  • Master + Policy (RAG) + Ticket (tool-using) agents
  • Evaluated with 15 scenarios and an LLM-as-a-judge
  • Open source, with an 8-part write-up
  • Also rebuilt in LangGraph
Google GeminiRAGPythonFastAPI
View Details →

Fashion Recommendation System

Seedcom R&D | Nov 2020 - Oct 2024

Led team of 3 engineers to build computer vision solution for fashion retail using RetinaNet and OpenCV.

  • 92% average precision in outfit detection
  • Led cross-functional team of 3 engineers
  • Increased customer click-rate and cross-sell
  • Deployed across web and mobile platforms
RetinaNetOpenCVDockerFastAPI
View Details →

Semiconductor Defect Detection

Emage Development | Jul 2019 - Jul 2020

Production-ready defect detection system for semiconductor manufacturing quality control.

  • 98% precision in defect identification
  • 0.2-second inspection time per image
  • Real-time factory floor deployment
  • Complete ML workflow with WPF application
YOLOv3RetinaNetOpenCVC#/WPF
View Details →

Vietnamese Recipe Chatbot with RAG

Seedcom R&D | Nov 2020 - Oct 2024

Intelligent chatbot using Retrieval-Augmented Generation with Google Gemini for personalized recipe recommendations.

  • RAG system with customer basket history
  • Personalized culinary suggestions
  • Enhanced user engagement metrics
  • FastAPI deployment for high-throughput
Google GeminiRAGPythonFastAPI
View Details →

ML Pipelines on Google Vertex AI

Seedcom R&D | Nov 2020 - Oct 2024

End-to-end automated ML pipelines for competitor price mapping and product recommendations.

  • CI/CD with GitLab CI and Airflow
  • Comprehensive monitoring and deployment
  • Scalable production infrastructure
  • Real-time analytics integration
Vertex AIGitLab CIAirflowGCP
View Details →

Parkinson's Disease Treatment Research

Northumbria University | Jan 2023 - Jul 2023

Predictive ML models for evaluating wearable/smartphone cues effectiveness in Parkinson's drooling treatment.

  • 65% F1-score in treatment schedule prediction
  • Deployed on AWS SageMaker
  • Clinical data analysis and visualization
  • Research contribution to medical science
Random ForestsAWS SageMakerPythonHealthcare AI
View Details →

About twentytwotensors

I'm Truong Giang Nguyen, founder of twentytwotensors — a London-based AI and machine learning consultancy. I have an MSc in Data Science (Distinction) and 5+ years building production ML: a semiconductor defect-detection system running at 98% precision and 0.2s an image, and a fashion recommendation engine at 92% precision, where I led a team of three.

My technical expertise spans natural language processing, computer vision, and cloud-native ML infrastructure. I've delivered solutions across manufacturing, retail, and healthcare — from enterprise LLM integration to real-time defect detection systems.

I bring end-to-end capability: data engineering, model development, deployment, and ongoing optimisation on GCP and AWS platforms.

15+ Projects Delivered
3 Industries Served
MSc Data Science (Distinction)
GCP Certified
Truong Giang Nguyen, AI/ML consultant and founder of twentytwotensors
MSc Data Science
with Distinction
Good Clinical Practice
Certified 2024
London, UK
Based
5+ Years
Production Experience

Academic & Industry Experience

MSc Data Science (Distinction) combined with 5+ years deploying production ML systems across enterprise environments

UK-Based Consultant

London-based for on-site collaboration. Full understanding of UK data protection, GDPR compliance, and business requirements

Production-Proven Results

Documented outcomes: 98% model accuracy and 92% recommendation precision in live systems

End-to-End ML Delivery

Complete machine learning lifecycle: data engineering, model development, cloud deployment, monitoring, and maintenance

Latest from the Blog

Technical insights on AI/ML engineering and deployment

June 2026

What does a factory's electricity actually cost?

A data project: estimate a GB factory's day-ahead wholesale electricity exposure (~10.8 p/kWh), then prove it against what actually traded — within 0.5%. On scoping, honest uncertainty, and validation.

Read more →
June 2026

A leakage-free fraud feature store, built with dbt

Turning raw card-payment tables into a feature store for fraud models on dbt and BigQuery — with one rule that matters: a feature can only ever see data from before the transaction it describes.

Read more →
June 2026

MedGemma CKD: building a clinical RAG assistant three ways

Why I built the same chronic-kidney-disease assistant at three levels — simple, agentic, multi-agent — and the shared foundation underneath. Part 1 of the MedGemma CKD series.

Read more →
June 2026

Evaluating a clinical RAG — the honest results

RAGAS, a retriever comparison, and middling scores I'm not going to dress up. Why retrieval over dense clinical text is the hard part. Part 6 of the MedGemma CKD series.

Read more →
June 2026

I built a system to run my life around a chronic illness

A working app I built for myself — it logs my treatments, reads my own blood-test trends, and answers questions with an AI assistant. The start of a build-in-public series.

Read more →
June 2026

Treating at home saves the NHS thousands a year

The cost case for treating a chronic illness at home: ~£9–12k saved per patient per year, an NHS target the country is missing, and why better software closes the gap.

Read more →
June 2026

An AI assistant that reads my actual health data

A function-calling assistant whose system prompt is rebuilt every turn from my live data, with symptom-driven retrieval and voice input. A reusable pattern for AI over your own data.

Read more →
June 2026

The readiness score: what I'm building next

The honest roadmap: a live body-data foundation, the daily readiness score I'm building on top of it, and why the pattern fits any fluctuating-capacity condition.

Read more →
June 2026

ADK vs LangGraph: one AI support system, two frameworks

I built the same production customer-support agent twice — on Google ADK and on LangGraph. Same model and data; only the orchestration changed. What differs, and how to choose.

Read more →
August 2024

I built a customer support agent as a team, not one bot

One coordinator, two specialists, and why I didn't build it as one big agent. The architecture the rest of the series sits on.

Read more →
December 2024

Computer Vision in Manufacturing

How computer vision is transforming quality control and defect detection in manufacturing.

Read more →
December 2024

MLOps Best Practices

Essential MLOps patterns for deploying and maintaining ML systems in production.

Read more →

Start Your AI Project

Ready to explore how machine learning can transform your business? Schedule a free consultation to discuss your requirements, technical approach, and project scope.

  • contact@twentytwotensors.co.uk
  • +44 79 0403 5063
  • London, United Kingdom
  • Response within 24 hours

Request Consultation