ML Engineering · NLP Research · Causal Inference

Humayun Kayesh

Humayun Kayesh -
ML Engineer & NLP Researcher

I architect production AI systems that scale - from an ML-based marketing platform built from scratch, to AI content and image generation engines, to published research on causal inference in social media. PhD, Commonwealth Scholar, 10+ years across the full path from research to deployment.

View projects Download résumé
PhD
ML & NLP, Griffith University
Commonwealth Scholar
University of Manchester
10+ yrs
Software, data & ML
236
Citations · h-index 10
  • Griffith University
  • University of Manchester
  • Robotic Marketer
  • HealthcareLogic
  • DeepCognito

Selected Work

Projects with Measurable Outcomes

Kaggle · Live demo

Disaster Tweets Classification

During a crisis, disaster-related tweets drown in noise. Responders need them filtered automatically - but short, unstructured text makes that a hard classification problem.

  • DistilBERT
  • Transfer learning
  • Transformers
  • Python

83%+ Test accuracy · live demo on Hugging Face

Code Demo

NLP research · NHS (UK)

Patient Opinion Mining from Feedback Comments

NHS trusts collect thousands of free-text patient comments but had no scalable way to tell which aspects of care - staff attitude, waiting time, environment - patients were praising or complaining about.

  • Machine learning
  • NLP
  • n-gram features
  • Topic classification
  • R

NHS Study Corpus & methods published in Health and Social Care Delivery Research

Code

Production AI · Robotic Marketer

AI-Powered Marketing Platform

A direct marketing platform needed to generate on-brand, customised content at volume - keyword and manual workflows couldn't keep up.

  • GPT fine-tuning
  • Claude / OpenRouter / DeepSeek
  • OpenAI embeddings
  • Flask
  • Redis Queue
  • AWS Lambda
  • Django

ML Architecture ML marketing platform architected end-to-end

Proprietary - details on request

Healthcare ML · HealthcareLogic

Hospital Length-of-Stay Prediction

Hospitals plan beds, staffing and resources around how long patients will stay - but the existing length-of-stay model wasn't accurate enough to base those decisions on.

  • Python
  • scikit-learn
  • Keras
  • EHR data
  • Feature engineering

+47% Improvement over the existing LoS model

Proprietary - details on request

Browse all projects

Expertise

A Full-stack Path from Research to Production

LLMs & Gen AI

  • GPT fine-tuning
  • OpenAI / Claude / OpenRouter / DeepSeek APIs
  • OpenAI embeddings & semantic similarity
  • AI content generation
  • Image generation
  • Transfer learning (BERT / DistilBERT)
  • Topic modeling
  • Sentiment analysis
  • Causality detection

MLOps

  • Continuous learning frameworks
  • Model evaluation & experiment design
  • CI/CD (CodeCommit, CodeDeploy)
  • Serverless pipelines

Cloud & Serverless

  • AWS Lambda
  • API Gateway
  • EC2
  • DynamoDB
  • S3
  • SQS
  • Redis Queue

Data Engineering

  • Python (pandas, NumPy, matplotlib)
  • scikit-learn
  • TensorFlow / PyTorch
  • ElasticSearch / OpenSearch
  • Scraping pipelines
  • R

Full-Stack Development

  • Python
  • Django
  • Flask APIs
  • ReactJS
  • JavaScript
  • AngularJS / Ionic
  • PHP
  • Drupal
  • RASA
  • Agile delivery

Database

  • MongoDB
  • PostgreSQL
  • Database design
  • ORM / ODM

Experience

Where I've Made an Impact

Jun 2022 - Present

Machine Learning Engineer

Robotic Marketer Ltd. · Melbourne, Australia

  • Architected an ML-based direct marketing platform from scratch, handling thousands of data points.
  • Developed an AI-powered content generation engine and exposed it through a REST API.
  • Created an image generation platform that lets users reliably generate brand-specific content.
  • Designed and supervised the development of a CRM system featuring lead pipeline tracking and management.
  • Built serverless applications to automate website scraping, data extraction and analysis.

ML-based marketing platform architected from scratch

Jul 2021 - Apr 2022

Research Assistant (part-time)

HealthcareLogic Pty Ltd. · Gold Coast, Australia

  • Designed and developed a machine learning model for hospital length-of-stay (LoS) prediction.
  • Experimented, analysed and evaluated several machine learning models for regression.
  • Developed a patient arrival rate prediction model for hospitals.

LoS model outperformed the existing one by at least 47%

Jun 2020 - Sep 2020

AI Research Engineer (part-time)

DeepCognito Ltd. · Manchester, UK

  • Designed, implemented and evaluated the performance of several machine learning models for continuous learning.
  • Implemented several core components of an existing continuous learning framework in Python.
  • Designed and implemented a database similarity measurement approach to improve data retrieval and processing accuracy.

Database similarity approach improving retrieval accuracy

Oct 2017 - Mar 2018

Software Engineer (part-time)

DeepCognito Ltd. · Manchester, UK

  • Developed a conversational AI chatbot using RASA and ML that answers user queries.
  • Extracted topics and associated sentiments from patient feedback comments.
  • Completed an ML model to detect sentiment in news articles.

Conversational AI chatbot built with RASA

Dec 2015 - Aug 2016

Software Engineer

Jaxara IT Limited · Dhaka, Bangladesh

  • Transformed client requirements into requirement specifications.
  • Designed and set up system architecture by following standard practices.
  • Applied tools such as Drupal, Jira and Git for implementation and documentation.

Jun 2013 - Mar 2015

Software Engineer

Nascenia IT · Dhaka, Bangladesh

  • Developed a betting-related PHP web application requiring REST API integration.
  • Integrated the social media login API into a web application.

Research

Causal Inference & Social Media NLP

My research asks how machines can detect why things happen - finding causal relationships buried in tweets and patient feedback using deep learning and transfer learning.

arXiv:1901.03526 · 2019

On event causality detection in tweets

A neural approach to identifying causal relationships between events mentioned in tweets - the foundation of my PhD work on causal discovery in text.

  • Causal inference
  • Social-media NLP
  • Deep learning

31 citations Read paper

236 citations · h-index 10 · i10-index 10 Full Google Scholar profile

Boga Lake in the Bandarban hill district of Bangladesh, surrounded by forested mountains

Beyond Work

Off the Beaten Path

Away from the keyboard I hike - mountains, lakes, beaches, forests. The benchmark so far: a 2015 trek to Boga Lake, a remote mountaintop lake in Bandarban, Bangladesh.

Contact

Have an AI problem worth solving?

I'm happiest where research rigour meets production reality - LLM features, NLP pipelines, search & recommendation, or making sense of messy text data.

hkayesh@gmail.com LinkedIn