Ming Ze Tang is an AI and full stack software engineer based in Belfast, UK and Kuala Lumpur, Malaysia. He builds scalable web platforms and high-throughput backend systems, and evaluates LLM systems under rigorous benchmarking frameworks.

Does Ming Ze Tang require visa sponsorship to work in the UK?

No. Ming Ze Tang holds a UK Global Talent Visa with full right to work in the UK. No visa sponsorship is required. He is open to roles in the UK and Malaysia.

What are your primary areas of technical expertise?

Ming Ze Tang works with Python (PyTorch, TensorFlow, Hugging Face) for AI and NLP systems, Java (Spring Boot) and Apache Kafka for high-throughput distributed architectures, and TypeScript, React, and Next.js for modern user interfaces.

What research have you published in machine learning?

Ming Ze Tang has published research on benchmarking temporal biases in large language models (DateLogicQA, NAACL 2025 Student Research Workshop) and on zero-shot multimodal posture classification under data scarcity (MMLoSo 2025, an ACL workshop).

Where are you currently based and do you work remotely?

Ming Ze Tang works across Belfast, UK and Kuala Lumpur, Malaysia, and is fully equipped for remote collaboration across time zones.

whoami.txt

operator profile

Ming Ze Tang

AI and Full Stack Software Engineer

Belfast, UK / Kuala Lumpur, MY

UK GLOBAL TALENT VISA / NO SPONSORSHIP REQUIRED

open to work: uk and malaysia

Right to work in the UK. UK Global Talent Visa holder. No visa sponsorship required.

download cv email me

Who I am

I am an AI and full stack software engineer working in Belfast, UK, originally from Kuala Lumpur, Malaysia. Over the past three years I have shipped production software ranging from AI screening pipelines to distributed backend systems that sustain high-volume traffic.

My focus is the gap between machine learning and reliable systems engineering. I do not treat models as black boxes. I design pipelines with measurable behaviour, transparent results, and performance budgets that hold up in production.

What I build

AI and machine learning systems with production rigor
Distributed backend infrastructure and microservices
Full stack web applications with React and TypeScript
Research on LLM evaluation and AI system benchmarking
High-performance data processing pipelines
Cloud-native applications on Kubernetes

Measured outcomes

metric	result	how
query performance	10s to <1s	Optimized API and database paths for sub-second retrieval.
message throughput	5000+/hour	Sustained ingestion and processing at production scale.
payroll operations time	40% less	Automated cross-functional ERP workflows with system-level controls.
manual workflow steps	70% fewer	AI-assisted pipelines that remove repetitive human tasks.

experience.log

selected production work

[ai-screening]AI Screening Pipeline for SEIS/EIS Evaluation

Built an AI-driven screening workflow to evaluate SEIS/EIS company profiles with structured scoring and explainable ranking outputs.

Python / NLP / Data Pipelines / Model Evaluation

impact: Improved analyst throughput while increasing consistency and decision transparency.

[legal-nlp]Legal Knowledge Management with NLP Recommendations

Developed an AI-powered legal intelligence platform that indexed domain knowledge and recommended relevant references using semantic search.

LLMs / Vector Search / TypeScript / Backend APIs

impact: Reduced retrieval friction and enabled faster legal insight workflows for knowledge users.

[erp]ERP Platform for HR, Sales, and Warehouse Operations

Engineered core ERP modules and integrations to unify operations across departments, with robust role controls and auditability.

Java / Spring Boot / SQL / Angular

impact: Delivered measurable operational gains including a 40% payroll process time reduction.

[logistics-crm]Real-Time Logistics and CRM Performance Engineering

Implemented real-time data flows and optimized backend processing paths for logistics and CRM workloads under sustained traffic.

Apache Kafka / Distributed Systems / Observability / API Optimization

impact: Achieved 10x query latency improvements and supported 5000+ processed messages per hour.

term: mzsh

mzsh on mzwm 1.0.4. type help to list commands.

guest@mzwm:~$ help

available commands:

whoami name, role, location, visa status

skills the stack, grouped

experience selected production work

publications peer-reviewed papers

contact email and profiles

cv download the pdf resume

man tang the manual, with faq

ls list open windows

uptime session statistics

clear clear the screen

tab completes. up arrow recalls history.

guest@mzwm:~$

publications.bib

peer-reviewed research

DateLogicQA: Benchmarking Temporal Biases in Large Language Models

NAACL 2025, Student Research Workshop / llm evaluation, temporal reasoning, benchmark design

Introduces a benchmark for temporal reasoning failure modes in LLM systems and analyzes model behaviour under structured evaluation settings.

bibtex

@inproceedings{bhatia-etal-2025-datelogicqa,
  title = "{D}ate{L}ogic{QA}: Benchmarking Temporal Biases in Large Language Models",
  author = "Bhatia, Gagan and Tang, MingZe and Mahanta, Cristina and
            Kazi, Madiha and Peyrard, Maxime and Zhao, Wei",
  booktitle = "Proceedings of the 2025 Conference of the Nations of the Americas
               Chapter of the Association for Computational Linguistics: Human Language
               Technologies (Volume 4: Student Research Workshop)",
  month = apr,
  year = "2025",
  address = "Albuquerque, USA",
  publisher = "Association for Computational Linguistics",
  url = "https://aclanthology.org/2025.naacl-srw.32/",
  doi = "10.18653/v1/2025.naacl-srw.32",
  pages = "321--332"
}

Language as a Label: Zero-Shot Multimodal Classification of Everyday Postures under Data Scarcity

MMLoSo 2025 (ACL workshop) / computer vision, vision language models, zero-shot classification

Investigates how the specificity of natural language prompts influences zero-shot classification performance in vision language models under severe data scarcity.

bibtex

@inproceedings{tang-jacob-2025-language,
  title = "Language as a Label: Zero-Shot Multimodal Classification of Everyday
           Postures under Data Scarcity",
  author = "Tang, MingZe and Jacob, Jubal Chandy",
  booktitle = "Proceedings of the 1st Workshop on Multimodal Models for
               Low-Resource Contexts and Social Impact (MMLoSo 2025)",
  month = dec,
  year = "2025",
  address = "Mumbai, India",
  publisher = "Association for Computational Linguistics",
  url = "https://aclanthology.org/2025.mmloso-1.5/",
  pages = "48--57"
}

view all research on google scholar

skills.tbl

group	stack
ml	python, pytorch, tensorflow, nlp, computer vision, llm evaluation
backend	java, spring boot, apache kafka, sql, performance tuning, system design
frontend	typescript, react, next.js, angular, api design, product delivery
infra	docker, kubernetes, ci/cd, observability, cloud services, scalable deployment

no bars, no percentages. ask me about any of these in an interview.

man tang(1)

TANG(1)mzwm user commandsTANG(1)

name

tang - AI and full stack software engineer, Belfast UK / Kuala Lumpur MY

synopsis

tang [--ai] [--fullstack] [--no-sponsorship-required]

location

Works across Belfast, United Kingdom and Kuala Lumpur, Malaysia. Remote collaboration across time zones is standard practice.

visa

Right to work in the UK. UK Global Talent Visa holder. No visa sponsorship required.

questions

Who is Ming Ze Tang?: I am an AI and full stack software engineer based in Belfast, UK and Kuala Lumpur, Malaysia. I build scalable web platforms and high-throughput backend systems, and I evaluate LLM systems under rigorous benchmarking frameworks.
Does Ming Ze Tang require visa sponsorship to work in the UK?: No. I hold a UK Global Talent Visa with full right to work in the UK. No visa sponsorship is required. I am open to roles in the UK and Malaysia.
What are your primary areas of technical expertise?: My skills span machine learning and full stack software development. I work with Python (PyTorch, TensorFlow, Hugging Face) for AI and NLP systems, Java (Spring Boot) and Apache Kafka for high-throughput distributed architectures, and TypeScript, React, and Next.js for modern user interfaces.
What research have you published in machine learning?: I have published research on benchmarking temporal biases in large language models (DateLogicQA, NAACL 2025 Student Research Workshop) and on zero-shot multimodal posture classification under data scarcity (MMLoSo 2025, an ACL workshop).
Where are you currently based and do you work remotely?: I work across Belfast, UK and Kuala Lumpur, Malaysia. I am fully equipped for remote collaboration across time zones, with hands-on experience building systems that handle substantial traffic and transaction volumes.

bugs

Occasionally optimistic about estimates. Report to business@mingzetang.dev.

contact.vcf

email	business@mingzetang.dev
github	github.com/tangmz
linkedin	linkedin.com/in/ming-ze-tang
medium	medium.com/@mztang0428
scholar	scholar.google.com

send an email

I read and respond to every message.

drafts.d

ideas removed from this repository, by version tag

v1.0.0: hero with a 3d robot.
deleted: shipping three.js to say hello was not worth it.
v1.0.2: scroll-based storytelling.
deleted: nobody scrolls a cv for fun.
v1.0.4: glassmorphism cards with glow effects.
deleted: every portfolio has them. that is the problem.

nothing else in here. the good ideas shipped.

lab

nothing running here yet. this workspace is reserved for experiments.

head back to workspace 1, or type man tang in the terminal.

MingZeTang_CV.pdf linkedin

framework	next 16.2.1, react 19.2.4
fonts	manrope + jetbrains mono, self hosted via next/font
rendering	tiled, server rendered, readable without javascript
shortcuts	alt+1..3 workspaces, alt+j/k cycle focus

Ming Ze Tang

Who I am

What I build

Measured outcomes

Experience highlights

[ai-screening]AI Screening Pipeline for SEIS/EIS Evaluation

[legal-nlp]Legal Knowledge Management with NLP Recommendations

[erp]ERP Platform for HR, Sales, and Warehouse Operations

[logistics-crm]Real-Time Logistics and CRM Performance Engineering

Research publications

DateLogicQA: Benchmarking Temporal Biases in Large Language Models

Language as a Label: Zero-Shot Multimodal Classification of Everyday Postures under Data Scarcity

Technical skills

Frequently asked questions

name

synopsis

location

visa

questions

see also

bugs

Contact Ming Ze Tang

Deleted drafts

Lab workspace