S M Hasan Download CV

Hasan

S M Mahamudul Hasan, Applied AI Engineer based in London

See my projects Email me

Right now

  • Applied AI Engineer at BritAI Labs since January 2026
  • MSc in Artificial Intelligence and Robotics, University of Hertfordshire, 2024 to 2026
  • Open to Applied AI, machine learning and GenAI engineering roles in London

About

I'm an Applied AI Engineer with more than three years across enterprise software, pricing platforms and AI services. I like the part of the work where a model has to become something people can actually run.

At BritAI Labs I built a private bilingual chat service on a local LLM with FastAPI, Ollama and Docker Compose, and a job-placement model that reached 85% accuracy. Before that I spent two years configuring Pricefx pricing systems in Groovy for B2B manufacturers, where SAP integration cut manual pricing work by 95%.

My research started at HAB Lab in Sylhet, where I led a five-member team and co-authored papers with IEEE and Springer Nature on Bangla fake-news detection, phishing detection and cancer prediction.

LLM applications
Local models, streaming APIs and chat interfaces with FastAPI, Ollama and Docker.
Bangla language technology
Fake-news detection, text classification and bilingual products.
Applied machine learning
Health and security prediction models, benchmarked and published.
Robotics and simulation
Sensor-driven robots in Unity and agent-based models in NetLogo.
Enterprise pricing systems
Pricefx, Groovy, SAP integration and CPQ for B2B clients.

Selected projects

Five projects that show how I work, from a local LLM service to a self-driving robot. Each one links to its code.

LLM application

Local LLM Chat API

A private chat service that runs on a local model and replies in Bangla or English.

  • Streams replies token by token as NDJSON, or returns one JSON response
  • Keeps the last 20 messages of each session as context
  • Exports conversations as JSON or Markdown
  • Runs with one Docker Compose command, with health checks and unit and API tests that don't need Ollama
  • Python
  • FastAPI
  • Ollama
  • Docker Compose
  • pytest
View code
Docker Compose
Chat UIPlain HTML, CSS and JavaScript in the browser
JSON or NDJSON stream
FastAPI serviceSession memory, Bangla and English prompts, export
HTTP
Ollamallama3.2:3b by default, any installed model
Main endpoints
POST/api/chatFull JSON reply
POST/api/chat/streamToken stream
GET/api/sessions/{id}/exportJSON or Markdown
GET/health/readyModel readiness
Bilingual web app

RoadReady

An English and Bangla study app for the GB car theory test, live on GitHub Pages.

  • 648 bilingual practice questions and 165 learning rules
  • 50-question, 57-minute mock tests with a 43 out of 50 pass mark
  • Shows English, Bangla or both side by side
  • Saves progress, streaks and mistakes in the browser, with no account needed
  • TypeScript
  • Vite
  • GitHub Actions
  • GitHub Pages
Open the live app View code

Each dot is one practice question, banded by topic. Pick a topic to highlight it.

Bangla NLP research

Bangla Fake News Detection

A full pipeline for spotting fake news in Bangla, from dataset to mobile app. Published with IEEE.

  • Benchmarks classical ML, LSTM, RNN and Bangla BERT models
  • The RNN reached 96.55% accuracy; Bangla BERT reached an F1 score of 0.96
  • Predictions served through a Flask API to a Flutter mobile app
  • The work was adopted by a media agency
  • Python
  • TensorFlow
  • Keras
  • scikit-learn
  • Bangla BERT
  • Flask
  • Flutter
View code Read the IEEE paper
  1. 11,434Bangla news articles collected and labelled as fake or authentic
  2. 5models benchmarked, from TF-IDF with SGD to Bangla BERT
  3. 96.55%accuracy from the best model, a recurrent neural network
  4. APIthe trained model answers requests through Flask
  5. Appreaders check a story on their phone in Flutter
Robotics simulation

Autonomous Robot Simulation

A Unity robot that drives itself along a track, steering around obstacles and powering up slopes.

  • Raycast sensors detect obstacles, slopes and road edges
  • Steering angle and motor force adjust from the sensor readings
  • Uses Unity's Rigidbody and WheelCollider physics
  • Built for the Artificial Life with Robotics module of my MSc
  • Unity
  • C#
  • Physics simulation
View code
The robot drives down a grey track lined with red and white kerbs, recorded in Unity
Recorded in Unity. The robot keeps to the track using only its sensors.
Agent-based modelling

Disease Spread Simulation

A NetLogo model for testing how public-health measures change the way a disease spreads.

  • Two regions of agents, each with its own population and starting conditions
  • Travel restrictions, social distancing and self-isolation switch on or off mid-run
  • Live monitors for infections, deaths and immunity
  • Sliders for infection rate, survival rate, immunity and illness duration
  • NetLogo
  • Agent-based modelling
  • Policy simulation
View code
The NetLogo world: a pink region above a grey region, with infected agents in green spreading through both
A run with 2,500 agents: 1,500 in the pink region and 1,000 in the grey.
  • Travel restrictions
  • Social distancing
  • Self-isolation

Every repository

All 25 of my original projects across two GitHub accounts: hasansust32 for work and research, smhasan24 for MSc coursework. Each row opens the repository.

    Research

    Three peer-reviewed papers with IEEE and Springer Nature, plus a research paper on heart-attack prediction. Pick a paper to see the models it compared.

      Experience

      AI engineering, enterprise pricing systems and research, most recent first.

      1. Jan 2026 to nowLondon, fixed-term contract

        Applied AI Engineer

        BritAI Labs

        • Built a private bilingual chat service on a local LLM with FastAPI, Ollama and Docker Compose, including NDJSON streaming, a 20-message context window, health checks, export and unit and API tests.
        • Reached 85% job-placement prediction accuracy with an R model using forecasting and statistical analysis.
        • Built a NetLogo agent-based simulation to compare three disease-intervention strategies.
      2. Jun 2024 to Sep 2024Contract, remote, United States

        Application Engineer

        All Generation Tech

        • Built and shipped backend applications on the Unqork no-code platform, cutting deployment time by 30%.
        • Integrated APIs and JSON data flows, improving system efficiency by 20%.
        • Scoped and delivered client change requests, reducing turnaround time by 40%.
      3. Nov 2023 to Apr 2024Contract, remote, United States

        Pricefx Configuration Engineer

        Eglobal Infotech, for client Canidium

        • Configured Pricefx pricing models in Groovy and integrated SAP source data, reducing manual pricing effort by 95%.
        • Automated data-source integrations and workflows for enterprise pricing operations.
      4. Jun 2022 to Sep 2023Full-time, remote, Australia

        Pricing Software Engineer

        Ark Solutions

        • Delivered three end-to-end B2B Pricefx implementations across three countries, from requirements to CPQ workflows.
        • Cut pricing-processing time by 25% by optimising Groovy rules with object-oriented design.
        • Earned the Pricefx Certified Configuration Engineer credential.
      5. Feb 2020 to May 2022Sylhet, on-site

        Machine Learning and NLP Research Apprentice

        HAB Lab (HCI, AI and Blockchain Lab)

        • Led a five-member research team and co-authored three papers published by IEEE and Springer.
        • Built and evaluated ML, deep-learning and NLP models for classification, sentiment analysis and information extraction.
      6. Sep 2018 to Feb 2020Part-time, remote

        WordPress Web Developer

        InnovexIT

        • Designed and launched more than 15 client websites, from e-commerce to corporate portfolios.
        • Improved site speed and SEO, raising organic traffic by 40%.

      Skills

      AI and machine learning

      • Python
      • scikit-learn
      • TensorFlow
      • PyTorch
      • Keras
      • BERT
      • LSTM and RNN
      • XGBoost
      • Word2Vec and GloVe
      • Feature engineering
      • Hyperparameter tuning
      • Model evaluation

      LLM apps and APIs

      • FastAPI
      • Ollama
      • LLM integration
      • REST APIs
      • JSON and NDJSON streaming
      • Flask
      • Docker Compose
      • pytest

      Data and analytics

      • SQL
      • R
      • Pandas
      • NumPy
      • Forecasting
      • Statistical analysis
      • Matplotlib
      • Seaborn
      • Plotly
      • Power BI

      Robotics and simulation

      • Unity
      • C#
      • NetLogo
      • Agent-based modelling
      • Sensor-based navigation

      Enterprise and delivery

      • Pricefx
      • Groovy
      • SAP integration
      • SAP CPQ
      • Unqork
      • Java
      • Git and GitLab
      • CI/CD
      • Jira and Scrum
      • Azure DevOps

      Web and tooling

      • TypeScript
      • JavaScript
      • HTML and CSS
      • WordPress
      • Flutter
      • Claude Code
      • GitHub Copilot
      • n8n

      Education and recognition

      Education

      MSc Artificial Intelligence and Robotics, with work placement

      University of Hertfordshire, Hatfield, 2024 to 2026

      Modules include Artificial Life with Robotics, Neural Networks and Machine Learning, Theory and Practice of Artificial Intelligence, Responsible Technology and a Team Research and Development Project.

      BSc Computer Science and Engineering

      Shahjalal University of Science and Technology, Sylhet, 2017 to 2022

      Algorithms, software engineering, databases and AI, with a research focus on natural language processing.

      Awards and certificates

      Hiring for an AI role in London?

      I'm open to Applied AI, machine learning and GenAI engineering roles. Email is the quickest way to reach me.

      hasansust32@gmail.com
      LinkedIn hasansust32 smhasan24 YouTube +44 7448 652378