Skip to content

portfolio/init

Zhixiang Feng

Machine Learning Researcher and Software Engineer focused on retrieval systems

I work on neural reranking, retrieval evaluation, and practical software that makes model behaviour easier to inspect and improve.

research focus
RAG
BSc honours
First Class
current degree
MSc ACSE
zhixiangfeng@portfolio: ~/profile_probe.sh

$

Machine Learning Researcher and Software Engineer focused on retrieval systems

$

Machine Learning Researcher at Revax, improving neural reranking for UK tax RAG. Open to new AI, ML, and software engineering positions.

$

ai_path

python eval_reranker.py --slice hard_tax_queries

software_path

npm test && ship --target browser cloud

$

ready for review, refactor, deployment

git_log --timeline

Experience shaped around proof

education

02

MSc Applied Computational Science and Engineering

Imperial College London

09/2025 to 09/2026 expected

  • Studying computational methods for applied science and engineering problems
  • Building on a computing science background with stronger modelling and numerical foundations
  • Using the programme to sharpen ML, simulation, and software engineering judgement
MSccomputational scienceengineering

BSc Computing Science

University of Glasgow

09/2021 to 07/2025

  • Graduated with First Class Honours and a specialism in Data Management
  • Received the Undergraduate Excellence Scholarship from 2021 to 2025
  • Built a foundation across algorithms, data, software engineering, and applied AI projects
First Class Honoursdata managementscholarship

industry & research

03

Machine Learning Researcher

Revax

05/2026 to present

  • Working on UK tax RAG retrieval quality, with the current focus on improving the neural reranker
  • Analysing hard queries and reranker failures so model changes are tied to search behaviour
  • Connecting evaluation results to practical changes in tax-focused question answering
RAGneural rerankingevaluationPython

Assistant Demonstrator

School of Computing Science, University of Glasgow

09/2024 to 06/2025

  • Delivered tutorials and hands-on labs for Computer Programming, Further Programming, and Algorithmic Foundations
  • Helped students work through core programming concepts and practical exercises
  • Marked assessed exercises and gave feedback that students could act on
coachingprogrammingalgorithms

Full Stack Developer Intern

MESYI Technology Ltd.

06/2023 to 08/2023

  • Built staff and business management workflows for a commercial B2B SaaS platform using Java, Spring Boot, JavaScript, and PostgreSQL
  • Designed and implemented REST APIs, integrated Spark-based data processing, and documented service boundaries, request flows, and database relationships
  • Led requirements analysis and technology selection, produced ER diagrams, wireframes, and system architecture diagrams, and supported the product from specification to commercial use
Spring BootREST APIsApache SparkSystem Design

role_router

Browse the portfolio by signal

python eval_reranker.py --slice hard_tax_queries

My current work is neural reranking for UK tax RAG. My research background covers adversarial attacks on ranking systems, BERT based question answering, and synthetic clinical label generation.

RAGneural rerankinginformation retrievalBERTsynthetic dataevaluation

npm test && ship --target browser cloud

I have built web, mobile and desktop apps, browser extensions, and SaaS tooling. I care about clear permissions, data safety, CI, and interfaces that stay usable under real users.

RustJavaScriptDjangoChrome extensionsCIsystem design

selected_work

Projects that explain how I think

Selected work from retrieval research, applied ML, browser tooling, and full stack projects. The focus is on what was built, what was evaluated, and where the engineering judgement mattered.

grep --role [ai|software]

revax-tax-rag-reranker/
featured AI SWE

2026

UK Tax RAG Neural Reranker

Current ML research work at Revax, focused on improving the neural reranker behind a UK tax retrieval augmented generation workflow.

PythonRAGneural rerankingretrieval evaluationtax domain QA
  • Improving retrieved evidence quality before answer generation
  • Analysing hard tax queries and reranker failure cases
  • Turning evaluation findings into practical reranker changes
synthetic-epilepsy-labels/
featured AI

2025

Synthetic Epilepsy Labels Generation

AI for healthcare research on generating synthetic multi-label epilepsy datasets from a small set of clinical letters.

Pythonsynthetic dataclinical NLPmulti-label classificationevaluation
  • Investigating synthetic labels for scarce real-world medical text
  • Designing checks for label diversity, accuracy, and reliability
  • Targeting more than 15,000 synthetic cases from fewer than 100 real samples
adversarial-ranking-attacks/
featured AI

2024 to 2025

Adversarial Attacks on Ranking Systems

Research on how adversarial attacks affect neural and non-neural retrieval pipelines in search systems.

Pythonneural rankinginformation retrievaladversarial attacksranking evaluation
  • Implemented recent adversarial attack algorithms targeting ranking components
  • Simulated search pipelines that combine neural and non-neural retrieval
  • Analysed how attacks changed document order, relevance scores, and search behaviour
hawk-page-indexer/
featured SWE

2023 to 2024

Hawk Page Indexer

Chrome extension built in a University of Glasgow team project with Amazon.com, Inc. as client, combining a page indexer, markdown notebook, and to-do list.

Project page: Hawk - Page Indexer on the Chrome Web Store

JavaScriptNode.jsChrome ExtensionMiniSearchGitLab CI
  • Implemented local page indexing and search over selected web content
  • Used BM25F with MiniSearch fallback for ranked document retrieval
  • Supported JSON import and export for indexed page backups
  • Published as Hawk - Page Indexer on the Chrome Web Store
  • Presented a final product recognised by software engineers at Amazon Edinburgh
portfolio-website/
SWE

2026

This Portfolio

The site you are reading now. A fast static portfolio with a terminal inspired interface, client side section switching with horizontal slide animations, and a neural network background rendered on a canvas.

AstroTypeScriptTailwind CSSCloudflare Pages
  • Built a single page section deck so the whole site switches instantly with no full page reloads
  • Tuned for discovery with Open Graph cards, JSON-LD structured data, and a generated sitemap
  • Kept it accessible and quick with reduced motion support, keyboard navigation, and a no script fallback
rainbow-tensor/
SWE

2026

Rainbow Tensor

Colourful tensor operation visuals for NumPy style arrays, built for notebooks, docs, and teaching shape transformations. The project has around 100 GitHub stars.

PythonNumPyJupyterSVGscientific computing
  • Built notebook and documentation visuals for tensor shape transformations
  • Packaged Python helpers for NumPy style arrays and teaching workflows
  • Earned around 100 GitHub stars from the developer and scientific Python community
nlp-qa-system/
AI

2024

NLP Question Answering System

Question answering project using WikiQA, set similarity methods, term frequency vectors, and BERT based context vectors.

PythonBERTWikiQAcosine similarityclassification metrics
  • Compared overlap coefficient, Sorensen-Dice, and Jaccard similarity
  • Evaluated BERT based answer selection with accuracy, precision, recall, and F1
  • Analysed model stability across different question and answer scenarios
autogether/
SWE

2023

AuTogether Event Registration App

Mobile event registration app for a non-profit organisation supporting children with autism and their families.

AngularIonicDjangoPythonJavaScript
  • Built event sign-up flows for parents and children
  • Added organiser tools for venue capacity and booking management
  • Connected mobile UI work with backend functionality
best-recipes/
SWE

2023

Best Recipes Web App

Recipe sharing web app where users could publish recipe blogs, rate posts, save favourites, and explore trends.

PythonSQLJavaScriptHTMLCSS
  • Worked on requirements analysis, UI design, technology selection, testing, and deployment
  • Built posting, editing, rating, and favourites functionality
  • Used the project to practise full stack delivery from schema to interface

git_tag --wins

Wins, and the people I helped along the way

Competitions worth shipping for, and the teaching and leadership roles where the goal was making other engineers better.

Hackathon wins

02

Weekend builds that shipped, got pitched, and placed first.

Imperial College Fintech Hackathon

2025

Imperial College London

Project Wompoo, a family-oriented product that protects seniors from phone-call fraud. We built and pitched it over a weekend against more than 100 students from universities across the UK and Europe, including KCL, UCL, Oxford, Southampton, and ETH Zürich, gathered to shape the future of finance and technology.

Built with Thien Phu Nguyen Xuan, Elaine Xu, Jorin Riexinger, and Polina Markova.

FintechFraud preventionProductPitch

MS Tanks, Do You Have the GUTS Hackathon

2024

Morgan Stanley

An autonomous tank agent built on a TCP API, an epsilon-greedy strategy, and a subsumption architecture, a simple behaviour hierarchy that let resource handling interrupt lower-priority actions as the game state changed.

Practical by design, it balanced exploration and exploitation rather than trying to be clever in every state.

Reinforcement learningAgent behaviourPython

Leadership & teaching

03

Helping newer engineers find their footing.

Class Representative

2025 to 2026

Imperial College London

Represented the cohort to staff, surfacing student feedback and helping steer how the course was delivered.

RepresentationFeedback

Peer-Assisted Learning Leader

2023 to 2024

University of Glasgow, School of Computing Science

Led two teams of new software engineers to design and build video games, guiding them from planning through implementation and delivery.

Scrum MasterMentoringGame dev

Mentor

2023 to 2025

University of Glasgow, School of Computing Science

Mentored junior Computing Science students across two years, supporting their coursework, study habits, and engineering practice.

Mentoring

capability_map

Tools are useful, but judgement is the real skill

01

Machine learning and IR

PyTorchscikit-learnHugging FaceRecommender SystemsRAGNeural Reranking
02

Programming

PythonRustCC++Parallel ProgrammingJavaScriptSQLJavaHaskellErlangGo
03

Backend and Apps

DjangoNode.jsTauriAxumChrome ExtensionsAngularIonicReact Native
04

Engineering Practice

GitCI WorkflowsCI/CDDevOpsTestingAgileCode ReviewSystem DiagramsCoaching
> Diagnose retrieval failures first, then validate every model change against concrete examples and measurable metrics
> Engineer software that is secure, observable, and maintainable, so any engineer can extend it with confidence under pressure

open_channel --secure

Send a focused message

Tell me what you are building, what is broken, or what role you are hiring for. A specific message beats a polished one.

$ mail contact@zhixiangfeng.com

$ pwd London, United Kingdom