Back to the doors

Research & Teaching

Research Assistant at Thammasat since 2019. Built LARICE@608 from the ground up and now directs it.

  • Three conference papers and an IEECON 2026 manuscript
  • Research Assistant (Engineer) since 2019
  • Built and directs LARICE@608, Thammasat
  • First-generation KidBright trainer
First pages of Top’s papers on blood-cell AI, solar panel defects and AI mobility, fanned out
Top presenting with a microphone, looking up at a projected slide of a drone
Presenting the research. IEECON 2026 manuscript · InCIT 2025 · EECON-43

Papers.

Three conference papers and a manuscript for IEECON 2026, from an autonomous scooter for the blind to real-time blood-cell analysis.

  1. First page of the manuscript on real-time white blood cell detection with YOLO11x-seg

    2025 · Manuscript · prepared for IEECON 2026

    AI-Based Real-Time WBC Detection and Classification from Peripheral Blood Smears Using YOLO11x-seg

    S. Tammawattana, P. Phutthammawong, E. Mukda and P. Kaewprapha

    Automated analyzers struggle with immature white blood cells, which sends labs back to slow manual microscopy. This YOLO11x-seg model segments and classifies ten cell types on peripheral blood smears. Trained with transfer learning on 7,882 public images, it reaches 0.963 [email protected] and runs at 25–30 FPS on an RTX 3080. It was compared with 10 medical technologists and aims to support resource-limited rural hospitals.

    • Instance segmentation
    • YOLO11x-seg
    • Hematology
    Explore it on LARICE Press
  2. Drone thermal imagery of a solar farm with detected hotspots and a table of panel temperatures

    2025 · IEEE InCIT 2025 · 9th International Conference on Information Technology · Phuket, pp. 43–49

    Enhancing Solar Panel Reliability: Cascade Mask R-CNN Defect Detection in Drone Thermal Images

    P. Phutthammawong and P. Kaewprapha

    Photovoltaic panel defects reduce energy generation and raise costs. In Thailand, manual inspections miss about 20% of defects, costing over US$433,000 per MW each year. The proposed Cascade Mask R-CNN system, trained on thermal images captured by drone, aims to cut inspection time from 11 hours to 90 minutes and could prevent up to US$2 million in annual losses for a 5 MW solar farm. Trained on more than 14,000 thermal images with a ResNet-101 backbone, the model reaches 77.3% [email protected]. Industry partner: B.Grimm Power.

    • Deep learning
    • Computer vision
    • Renewable energy
    Read on IEEE Xplore
  3. First page of the EECON-45 paper on NFT walletless identification

    2022 · EECON-45 · 45th Electrical Engineering Conference

    NFT Walletless Identification of Digital Assets and Physical Assets

    P. Phutthammawong, P. Kaewprapha and N. Somyat

    NFTs are moving beyond the purely digital and need to link to physical assets. The paper looks at the current limits, connecting digital ownership to real items and verifying who owns them, and proposes an NFT design that uses NFC tags for walletless identification, making physical–digital ownership simpler.

    • Blockchain
    • NFC
    • Digital identity
    Read on ResearchGate
  4. First page of the EECON-43 paper on an AI autonomous vehicle for the blind

    2020 · EECON-43 · 43rd Electrical Engineering Conference

    Artificial Intelligence Autonomous Vehicle for the Blind

    P. Phutthammawong, P. Angbunthorn, P. Kaewprapha and D. Cholaseuk

    An AI autonomous electric scooter for people with visual impairments. Artificial intelligence, SLAM and on-board sensors let it navigate safely on its own, opening the way to greater independence.

    • Artificial intelligence
    • SLAM
    • Assistive technology
    Read on ResearchGate

Built from the ground up.

LARICE@608 · Laboratory for Advanced Research, Innovation, Consultancy and Development in Computer Engineering, Electrical and Computer Engineering · Thammasat School of Engineering.

I built LARICE@608 from the ground up: the lab itself, its on-premise cloud and GPU fleet, the identity, CI/CD and monitoring underneath, and larice608.com.

Visit larice608.com
GPUs on-premise
10H100 · RTX 3080 · T4 · GTX 1060 ×7
GPU compute
~120 TFLOPSFP32 peak, published specs
Logical cores
180across 7 physical hosts
Memory
423 GiBacross 7 physical hosts
Kubernetes
3 nodesRKE2, etcd quorum of 3

Running on FreeIPA · GitLab CE · Jenkins · Kubernetes · Cloudflare Tunnel · OPNsense · Prometheus · Grafana

  1. Research

    Peer-reviewed work in AI, medical engineering, energy and sensing.

  2. Innovation

    Inventions and prototypes built for real users, from factory switchboards to nurse training.

  3. Consultancy

    Nine advisory services that turn industry problems into working solutions.

  4. Development

    Production systems the lab builds and runs: an on-premise cloud, CI/CD and a local LLM agent.

Five research areas

  • AI & computer vision
  • Medical engineering
  • Energy & IoT
  • Sensors & signal processing
  • Cloud, infrastructure & LLM systems

Nine consulting services

  • AI solutions
  • Data intelligence
  • Digital automation
  • Strategic consulting
  • PMO & delivery governance
  • Operations consulting
  • Software & hardware architecture
  • High-performance computing
  • 3D printing & prototyping

Publications & recognition from the lab

  • PaperEnhancing Solar Panel Reliability: Cascade Mask R-CNN Defect Detection in Drone Thermal ImagesIEEE InCIT 2025 · Phuket
  • ManuscriptAI-Based Real-time Blood Cell Detection and Classification from Peripheral Blood Smears using YOLO11x-segIEECON 2026
  • Award entryIoT Peak Demand Management and Carbon Credit Monitoring for SMEsThailand New Gen Inventors Award 2027 (I-New Gen Award)

More research.

Research and innovation projects since 2018, most of them leaving the lab as working systems.

Computer vision & AI in healthcare

  1. 2025A retrieval-augmented generation (RAG) LLM for personalised nutritional-supplement recommendations from health metrics and behavioural data
  2. 2020A microprocessor-controlled nocturnal orthodontic device with integrated bite-force monitoring
  3. 2019Automated foot-morphology classification from podoscope image processing
  4. 2018Machine-learning feature extraction and classification for early detection of diabetic retinopathy

Applied engineering & industrial solutions

  1. 2025A YOLO-guided computer-vision framework for colorimetric pH and chlorine measurement in water-quality testing
  2. 2019IoT-enabled data visualisation for industrial machinery (TSE TU-ROCKET) in manufacturing
  3. 2019Automated vehicle-damage assessment and severity classification with CNNs (VGG16)
  4. 2018A multi-campus IoT sensor network for PM2.5 monitoring: data acquisition and spatiotemporal analysis

Teaching.

From a lecture hall at Pathumthani to a lab of my own.

  1. Laboratory Director / Special Instructor; built the lab from the ground upLARICE@608, Laboratory for Advanced Research, Innovation, Consultancy and Development in Computer Engineering, Thammasat School of Engineering2025 – present
  2. Research Assistant (Engineer)Center of Academic Excellence in Integrative Advanced AI in Law, Engineering and Medicine, Thammasat2019 – present
  3. Special Invitation LecturerDepartment of Mechanical Engineering, Pathumthani University2019 – 2020
  4. Teacher AssistantFaculty of Engineering, Thammasat University
  5. Certified KidBright Trainer, first generationNational Electronics and Computer Technology Center (NECTEC)
Certificate from the Research and Innovation Division, Thammasat University, certifying that Promsurin Phutthammawong completed Principles of Ethics for Research Involving Humans on 30 May 2024

Certified.

  • Principles of Ethics for Research Involving HumansThammasat University, 30 May 2024 (valid to May 2026)
  • Human Subject Protection (HS)Valid 2025 – 2027
  • Good Clinical Practice, ICH-GCP E6(R2)Valid 2025 – 2027
  • Internal Quality Audit, ISO 14065:2020ISO standard training
  • Validation and Verification Bodies for Environmental Information, ISO/IEC 17029:2019ISO standard training
  • TOEICScore 785

Try another door.

Got a project?
Let’s build it together.

Research, a field system, a project that needs a PMO, or a wild moonshot. Send a message on LINE and I’ll get back to you.