{"schemaVersion":"1.0","provenance":"Owner-supplied professional and project descriptions; earlier non-conflicting portfolio material retained. Not independent verification.","profile":{"name":"Owen Cheung","role":"AI Engineer","tagline":"AI systems, built with judgement.","intro":"AI Engineer at LexisNexis Risk Solutions, building human-governed workflows for fraud-model optimisation. Final-year Computer Science & AI student at the University of Bath.","location":"Bath, UK","email":"oc608@bath.ac.uk","github":"https://github.com/OwenC05","linkedin":"https://www.linkedin.com/in/owen-cheung-472998225/","bio":["I'm Owen, an AI Engineer at LexisNexis Risk Solutions and a final-year Computer Science & AI student at the University of Bath. I stayed on as a contractor after my industrial placement to develop a self-initiated prototype into a funded internal pilot.","I care about systems that learn: representation learning, embeddings with the right geometry, and agentic systems that do real work. I’m heading toward AI research and engineering.","Away from the screen I snowboard with Bath Snowsports, pull a lot of espresso, and build mechanical keyboards — which is how I ended up typing at 193 wpm (and designing TypeForge)."]},"experience":[{"org":"LexisNexis Risk Solutions","role":"AI Engineer (Agentic AI)","period":"Aug 2026 — present · Contract","location":"London","note":"Human-governed fraud-model optimisation workflows; funded internal pilot building toward production."},{"org":"LexisNexis Risk Solutions","role":"Data Scientist","period":"Jul 2025 — Aug 2026 · Industrial placement","note":"Transaction analysis, two fraud-model optimisations, self-supervised representation learning and an agentic working demo."},{"org":"Dish'D","role":"Full-Stack Developer","period":"Jan 2024 — Feb 2025","location":"London","note":"Built a social cooking app end-to-end with Django + Flutter; led technical decisions."},{"org":"Kinetix","role":"System Tester","period":"Jul 2024 — Aug 2024","location":"Hong Kong","note":"QA for Census & Statistics Department systems — validated functionality against spec."},{"org":"DXC Technology","role":"AI Project Intern","period":"Jul 2022 — Aug 2022","location":"Hong Kong","note":"Drafted technical proposals for enterprise bids (AI-assisted mapping, OpenBIM.AI)."}],"education":[{"org":"University of Bath","detail":"BSc (Hons) Computer Science & Artificial Intelligence, with placement","period":"Sept 2023 — 2027 (expected)","note":"Expected 2:1"},{"org":"Wellington College","detail":"A-Levels — Maths, Further Maths, Physics, Computer Science","period":"Aug 2021 — Aug 2023"}],"skills":{"AI / ML":["PyTorch","Contrastive learning","Hyperbolic embeddings","Agentic systems","Deep learning"],"Languages":["Python","TypeScript","Haskell","Dart","SQL"],"Frameworks":["React / Node","Django","Flutter","Next.js"],"Data":["SQL","Snowflake","Statistical modelling","Data analysis"],"Cloud / AI":["Azure OpenAI","Azure Foundry and AWS S3 (workshop experience)"]},"interests":[{"label":"Snowboarding","note":"Carving and freestyle with Bath Snowsports — away from the screen."},{"label":"Specialty coffee","note":"Chasing brew methods and flavour."},{"label":"Mechanical keyboards","note":"Built several from scratch; reached 193 wpm."}],"projects":[{"id":"lexisnexis","slug":"lexisnexis-applied-ai","title":"LexisNexis — Applied AI","tagline":"Making fraud-model workflows recoverable and reviewable.","blurb":"A self-initiated agentic prototype, developed into a funded internal pilot with human review built in.","year":"Jul 2025 — present","role":"AI Engineer (Agentic AI) · previously Data Scientist","stack":["Python","SQLite","DVC","Papermill","LightGBM","Snowflake","PyTorch"],"summary":"During my placement I built a working demo for fraud-model optimisation. I was retained as an AI engineering contractor to develop the capital-funded internal pilot toward production.","highlights":["Custom Python orchestration, reusable Copilot skills, persistent SQLite state, audit trails and mandatory human review gates.","Recoverable DVC/Papermill workflows with LightGBM candidate-rule discovery and constrained logistic-regression policy weighting.","Placement work included two fraud-model optimisations and analysis of billions of transaction records using Python, SQL and Snowflake.","Researched self-supervised fraud representations using graph embeddings, contrastive learning and non-Euclidean geometry."],"status":"Internal pilot · toward production","context":"Fraud-model optimisation combines analysis, modelling and decisions that need accountable human review.","contribution":"Initiated a working agentic prototype during my placement, then stayed on as a contractor to develop the funded internal pilot.","approach":["Custom Python orchestration and reusable Copilot skills coordinate work with persistent SQLite state.","Audit trails and mandatory review gates keep human judgement in the workflow.","DVC and Papermill make notebook-based work recoverable; LightGBM discovers candidate rules, with constrained logistic regression for policy weighting."],"evidence":["Working demo delivered to the AI engineering team by the end of the placement.","Retained as an AI engineering contractor in August 2026 to develop a capital-funded internal pilot.","Two fraud-model optimisations completed during the earlier placement."],"limitations":["Employer-confidential work: no source code, transaction data or internal evaluation results are published.","This describes a pilot building toward production, not a launched production platform.","Descriptions are owner-supplied; no independently verified performance uplift is claimed."],"confidential":true,"selected":true,"url":"https://owencdev.info/projects/lexisnexis-applied-ai","markdownUrl":"https://owencdev.info/projects/lexisnexis-applied-ai/markdown"},{"id":"typeforge","slug":"typeforge","title":"TypeForge","tagline":"Practice that adapts to the way you type.","blurb":"A typing application in development, with implemented adaptive AI drills and real-time analytics being developed.","year":"Jul 2025 — present","role":"Solo developer","stack":["React","Node.js","TypeScript","Keystroke analytics"],"summary":"I am developing a typing application that tracks cadence, keystrokes and accuracy, with AI-driven drills designed and implemented to adapt to user weaknesses. It comes from building mechanical keyboards and reaching 193 wpm. A public demo is available; the product remains in development, with no public evaluation or full product release claimed.","highlights":["Designed and implemented adaptive AI-driven drills.","Developing real-time cadence, keystroke and accuracy analytics.","A personal exploration of more targeted typing practice."],"status":"In development","context":"Typing practice can focus on a headline speed rather than the particular weaknesses a person needs to practise.","contribution":"Designing and building a React/Node.js typing application, including implemented AI-driven adaptive drills.","approach":["Developing real-time analytics for cadence, keystrokes and accuracy.","Adaptive drills target user weaknesses.","Personal typing and keyboard-building experience informs the product direction."],"evidence":["Adaptive AI-driven drills designed and implemented, as described in my current project record."],"limitations":["A public demo is available; the product is still in development, not a full product release.","No measured learning uplift, timing precision or model benchmark is published."],"confidential":false,"selected":true,"url":"https://owencdev.info/projects/typeforge","markdownUrl":"https://owencdev.info/projects/typeforge/markdown","liveUrl":"https://typeforge-alpha.vercel.app"},{"id":"sortify","slug":"sortify","title":"Sortify","tagline":"Playlists sorted by feel.","blurb":"A Spotify tool that sorts and recommends playlists by genre, mood and danceability from audio features.","year":"2024","role":"Backend + recommendations","stack":["Python","Spotify API","Recommendation"],"summary":"A Spotify-API app that sorts and recommends playlists by genre, mood and danceability. I built the backend recommendation logic on top of Spotify's audio-feature data to surface personalised picks for how a playlist actually feels.","highlights":["Clusters tracks by audio features — mood, danceability, genre.","Personalised recommendations from listening signals.","Python backend against the Spotify Web API."],"status":"Earlier project","context":"Organising music around genre, mood and danceability.","contribution":"Built backend recommendation logic using Spotify audio features.","approach":["Python backend with the Spotify API.","Audio-feature-based playlist organisation and personalised recommendations."],"evidence":["Project summary retained from earlier portfolio work."],"limitations":["No public evaluation, repository or current live service is supplied.","This is historical project work, not a claim about current Spotify API availability."],"confidential":false,"selected":false,"url":"https://owencdev.info/projects/sortify","markdownUrl":"https://owencdev.info/projects/sortify/markdown"},{"id":"knowtrients","slug":"knowtrients","title":"Knowtrients","tagline":"Nutrition tracking, built by eight.","blurb":"A micronutrient + wellbeing app built in a team of 8 on a real Scrum cadence; I owned recipe parsing and nutrition analysis.","year":"Oct 2024 — May 2025","role":"Full-stack + Scrum","stack":["Django","Flutter / Dart","Agile / Scrum"],"summary":"A micronutrient-tracking and wellbeing app built in a team of 8 using Scrum. I implemented recipe parsing and nutrition analysis end-to-end with Django and Flutter, and helped run sprint planning and stakeholder testing — we onboarded 20+ pilot users.","highlights":["Recipe parsing + nutrition analysis, end to end.","Team of 8 on a real Scrum cadence.","20+ pilot users through stakeholder testing."],"status":"Team project · pilot tested","context":"A micronutrient-tracking and wellbeing app built by a team of eight.","contribution":"Implemented recipe parsing and nutrition analysis across Django and Flutter, and helped coordinate sprint planning and stakeholder testing.","approach":["Full-stack recipe parsing and nutrition analysis.","Agile Scrum collaboration and stakeholder testing."],"evidence":["Team of eight.","20+ pilot users through stakeholder testing."],"limitations":["Pilot participation is not a claim of retention or health outcomes.","No public repository or live deployment is supplied."],"confidential":false,"selected":true,"url":"https://owencdev.info/projects/knowtrients","markdownUrl":"https://owencdev.info/projects/knowtrients/markdown"},{"id":"dishd","slug":"dishd","title":"Dish'D","tagline":"A social network for cooking.","blurb":"A social cooking app (Django + Flutter) where I built auth, sharing and feeds, and led the technical decisions.","year":"Jan 2024 — Feb 2025","role":"Full-Stack Developer","stack":["Django","Flutter / Dart","Auth","Feeds"],"summary":"A social cooking app where people share recipes and meal inspiration. I built core features — authentication, content sharing and interactive feeds — with Django and Flutter, and led technical decisions on scalability, usability and performance.","highlights":["Auth, content sharing and interactive feeds.","Led technical decisions on scalability + performance.","Django backend, Flutter/Dart mobile client."],"status":"Previous product work","context":"A social cooking mobile application for recipes and meal inspiration.","contribution":"Developed core features and led technical decisions on scalability, usability and performance.","approach":["Django backend with Flutter/Dart mobile client.","Authentication, sharing and interactive feeds."],"evidence":["Full Stack Developer, January 2024 to February 2025."],"limitations":["No public adoption metrics, repository or live deployment is supplied."],"confidential":false,"selected":false,"url":"https://owencdev.info/projects/dishd","markdownUrl":"https://owencdev.info/projects/dishd/markdown"}]}