Lab

3 projects

All Projects

  • Customer Behaviour Analysis landing page with detection boxes drawn around figures in a collage of paintings

    Project · 2026 · Completed

    Customer Behavior Analysis with YOLOv26

    An end-to-end retail computer vision system that detects pre-purchase shelf behavior with a custom-trained YOLO model — mAP@50 60.91% across 4 interaction classes on 12,578 frames.

    • Python
    • Ultralytics YOLO
    • HuggingFace Transformers
    • Retail analytics
    • Dashboard
  • Eternora Analysis landing page with a hotel tower photo and the headline Welcome To Eternora Analysis

    Project · 2025 · Completed

    CiMSO Data Visualization Dashboard

    1st Place at the CiMSO Hospitality ERP Hackathon 2025 — a manager-centred analytics platform with 25 Chart.js visualizations, 45 KPIs, five languages and an LLM-powered insight assistant.

    • JavaScript
    • Node.js
    • Express
    • Chart.js
    • Dashboard
    • +4 more (listed on the detail page)
  • Sentiment trend chart with rolling sentiment lines for several SET tickers and dashed closing-price overlays

    Project · 2026 · Active

    Pantip SET Sentiment Monitor

    An automated NLP pipeline that scrapes Thai investor discussions from Pantip.com, links them to SET-listed tickers, scores sentiment with XLM-RoBERTa and serves a live Streamlit dashboard.

    • Python
    • Selenium
    • HuggingFace Transformers
    • XLM-RoBERTa
    • PyThaiNLP
    • +12 more (listed on the detail page)

All Papers

Papers I wrote

1

  1. Conference paper

    The Association Between Technology-Based Multitasking and Students' Academic Performance at Rangsit International College

    RSUCON 2026 research on how digital multitasking, cognitive load and technology use relate to student GPA. Python multiple linear regression (n=196) found Technology Use the sole significant predictor (β=0.137, p=0.004).

Papers I read

0

Written by other researchers — my notes, not my work.

Papers I study, with my notes, will be listed here.

Competition Analysis

  • Competition analysis · 2025 · 1st PlaceSample — placeholder content

    RIC Startup & Entrepreneurship Pitching Sessions: pitching LÉOR

    LÉOR, a wearable personal-safety concept — smart wristband and companion app — won 1st place; I shaped the product narrative, the Figma prototype and the live pitch.

    • Figma
    • Pitching
  • Team presenting on stage in front of a large screen showing the dashboard

    Competition analysis · 2025 · 1st Place of 11 finalist teamsSample — placeholder content

    CiMSO Hospitality ERP Hackathon 2025: deciding what not to show

    How a manager-first analytics dashboard — 25 charts, 45 KPIs, five languages and an LLM assistant — took 1st place out of 11 finalist teams.

    • JavaScript
    • Chart.js
    • Dashboard
    • LLM
    • Multilingual
    • +1 more (listed on the detail page)

This Week I Learned

  1. W422026

    Evaluating detection models Demo — placeholder content

    Week of

    • Concept

      mAP@50 and mAP@50–95 measure different things

      mAP@50 only asks whether a predicted box overlaps the truth by half; mAP@50–95 averages ten stricter overlap thresholds, so it rewards tight boxes. When localisation matters, report both.

      Object detectionModel evaluation

    • Observation

      Rare classes hide inside an average

      A macro-averaged score can look fine while one rare class is close to zero. Per-class precision and recall show where the model actually struggles.

      Model evaluation

  2. W412026

    Time series and sentiment Demo — placeholder content

    Week of

    • Concept

      A lagged correlation is not a forecast

      Correlation between sentiment and next-day returns only says the two move together after a delay. A Granger-style test or a hold-out period is needed before calling sentiment predictive.

      Time seriesStatistics

    • Learned

      Rolling z-scores adapt to slow drift

      Comparing each day with a rolling mean and standard deviation flags spikes without a fixed threshold that goes stale as volumes change.

      Anomaly detectionStatistics

    • Reading

      Notes on multilingual transformers

      Re-read how XLM-R was trained on text in 100 languages and why one shared vocabulary helps lower-resource languages such as Thai.

      XLM-RoBERTaMultilingualSource (opens in a new tab)

  3. W402026

    Small engineering lessons Demo — placeholder content

    Week of

    • Tool

      Scheduled GitHub Actions run in UTC, and can start late

      Cron schedules use UTC and may be delayed when runners are busy, so a pipeline should never assume it starts at an exact minute.

      GitHub Actions

    • Learned

      Tab panels can animate without a framework

      document.startViewTransition() snapshots the page, lets you swap the visible panel and animates between the two states with plain CSS.

      JavaScript

Contact

Send a message

Please double check your email address. Your details are only used to reply to you.

Or email me directly: contact@heinm.dev