Research Intern
Microsoft Core AI
Imperial College London MSc independent research, hosted with Microsoft Core AI. Building a local KYC/AML policy copilot with cited answers and a human-review boundary.
London / UK · Evidence-led engineering portfolio
Zhifeng Li
AI/ML Engineer & Computational Scientist
Building reliable AI systems, evidence-grounded RAG products and scientific computing workflows.
Select a project signal
About / 01
Engineering with evidence
I work where machine learning, numerical reasoning and dependable software systems meet.
I am an AI/ML engineer and computational scientist based in London, currently completing an MSc in Applied Computational Science and Engineering at Imperial College London, and a research intern with Microsoft Core AI.
I am most useful on problems where models must operate inside explicit constraints: local inference, traceable retrieval, reproducible experiments, numerical validity and honest evaluation. I prefer systems that expose their assumptions, fail safely and leave a clear path from claim to evidence.
Across current research and team coursework, I have worked with local RAG, computer vision, deep learning, genetic optimisation, scientific simulation and privacy-first recruitment email workflows. This portfolio separates completed work from prototypes and ongoing research so that every case study can be assessed on the evidence actually available.
Experience / 02
Current focus & education
The internship and research below are in progress. Academic credentials are limited to documented facts.
Now
On-Device RAG Copilot for KYC/AML
Research Intern, Microsoft Core AI
An in-progress Imperial MSc research project, undertaken as a research intern with Microsoft Core AI, building a local policy copilot for KYC/AML analysts with cited answers and a human-review boundary.
Read case studyEarlier experience
AI/NLP Development Engineer
EveryDay Learning · MHR × University of Nottingham
Worked on the AI study app's article-to-quiz pipeline, including question generation, distractor generation, seven-day spaced review scheduling, and learning-progress writes for the Android quiz experience.
Education
MSc Applied Computational Science and Engineering
Imperial College London
Applied AI, scientific computing, numerical methods and software engineering, including the current independent research project listed above.
BSc Computer Science
University of Nottingham
Undergraduate foundation in computer science and software development; degree classification recorded as First Class.
Capabilities / 03
Systems, models, evaluation
Capabilities are described through demonstrated project methods, without percentages or unsupported seniority claims.
- 01
Applied AI & evidence-grounded RAG
Local inference, hybrid retrieval, citation verification, abstention and human-review boundaries for policy-facing AI workflows.
Microsoft Core AI research internship · KYC/AML - 02
Scientific computing & optimisation
Simulation-aware optimisation, iterative numerical solvers, structural validity checks and controlled experimental comparisons.
Genetic Algorithm Circuit Optimizer coursework - 03
ML & data pipeline engineering
Reusable data, preprocessing, metrics, plotting and submission utilities supporting multi-task machine-learning experiments.
Storm Prediction ML team coursework - 04
Parallel & performance-aware computing
OpenMP-capable native tooling, seeded runtime comparisons and attention to the cost of validation and diagnostic checks.
C++20 circuit optimisation workflows - 05
Software engineering & evaluation
Typed boundaries, modular command-line workflows, reproducible tests and explicit separation between targets and measured outcomes.
Portfolio, vision prototype and research evaluation harnesses
Projects 01—06
Selected projects
Six evidence-led case studies across local AI, computer vision, enterprise AI/NLP product work, recruitment workflow agents, scientific machine learning and optimisation.
On-Device RAG Copilot for KYC/AML
An in-progress Imperial MSc research project, undertaken as a research intern with Microsoft Core AI, building a local policy copilot for KYC/AML analysts with cited answers and a human-review boundary.
- Python
- Microsoft Foundry Local
- Local SLM
- Hybrid RAG
- Sentence Transformers
ArUco + MediaPipe Hand Tracking
A computer-vision prototype combining calibrated ArUco pose estimation, MediaPipe hand landmarks and a Kalman-based fusion path with real-time visualisation and recordable experiment outputs.
- Python
- OpenCV
- MediaPipe
- NumPy
- Pandas
Recruitment Inbox Agent
An in-progress, privacy-first personal agent that turns campus-recruitment emails forwarded into a dedicated Outlook inbox into reviewable applications, calendar events and a daily brief, with the model limited to semantic extraction.
- Python
- FastAPI
- LangGraph
- LangChain
- Azure OpenAI
EveryDay Learning AI Study App
A shipped enterprise AI/NLP workflow for the Everyday Learning app that turned saved articles into quiz-ready questions, distractors and seven-day review sessions for the Android learning experience.
- Python
- PyTorch
- Hugging Face Transformers
- T5
- AWS Lambda
Predicting the Unpredictable
Team coursework exploring four storm-prediction tasks across 800 example storms, supported by a shared Python package for data, metrics, preprocessing, plotting and submission workflows.
- Python
- PyTorch
- scikit-learn
- Hugging Face
- NumPy
Genetic Algorithm Circuit Optimizer
A C++20 team coursework system combining circuit simulation, genetic optimisation, structural validity checks and reproducible seeded sweeps with optional robustness extensions.
- C++20
- CMake
- OpenMP
- Python
- Genetic algorithms
Contact / 05
Build something rigorous
I am interested in software engineering, AI/ML, applied research, scientific software and data-intensive engineering roles.
The public CV contains an email address and GitHub profile, with no phone number or private project identifier.