Rachapong Chirarattananon

Senior Full Stack Data Scientist at Codeboxx - Smart Digital Solutions
  • Claim this Profile
Contact Information
us****@****om
(386) 825-5501
Location
Bangkok City, Thailand, TH

Topline Score

Topline score feature will be out soon.

Bio

Generated by
Topline AI

You need to have a working account to view this content.
You need to have a working account to view this content.

Experience

    • Thailand
    • IT Services and IT Consulting
    • 1 - 100 Employee
    • Senior Full Stack Data Scientist
      • Jul 2023 - Present

      First task will probably be Marketing Attribution using Transformers. Will update some months into the role. First task will probably be Marketing Attribution using Transformers. Will update some months into the role.

    • Information Technology & Services
    • 100 - 200 Employee
    • Data Scientist
      • Mar 2022 - Oct 2022

      Set up and managed airflow docker deployment to host the team’s MLOps pipeline, writing DAGs, TaskGroups in a modular way that teammates can plug-in pipelines in the same DAG. The DAG template utilized dynamic variable names and functions such that the same class of tasks on different datasets are called by the same DAG code, with possible variation in procedures. Worked with millions, possibly trillions, transactions of structured data in a fraud detection task. Used both neural networks with Entity Embedding and tree-based models. Wrote an experimental Neural Architecture Search using local search algorithm. Keywords: Airflow, TaskGroup, DAG, ExternalTaskSensor, Tensorflow, Neural Network, Gradient Boosting, CatBoost, SHAP, Classification, Fraud detection, Big Data, SQL, MLFlow, MinIO, Docker Show less

    • Information Technology & Services
    • 1 - 100 Employee
    • Machine Learning Engineer
      • Jan 2021 - Feb 2022

      Developed instance segmentation model for applications on automobiles regarding claim-repair insurance process as part of the AI team. Headed the development of a specialized scene text recognition (OCR) model using DB-BiLSTM-CTC architecture. Both were developed from scratch, from data wrangling, monitoring the labeling, setting up training infrastructure, hp tuning, training, and evaluation. All was done on AWS cloud platform. Set up docker image application to be deployed by developer team. Keywords: AWS, S3, EC2, Sagemaker, boto3, Detectron2, Mmdetection, Mask-RCNN, CNN, QueryInst, Transformer, TorchVision, COCO, Instance Segmentation, PaddleOCR, OCR, LSTM, Docker, requests, Streamlit, LabelStudio Show less

    • Thailand
    • Food and Beverage Manufacturing
    • 700 & Above Employee
    • Data Scientist (Research Scientist)
      • Jul 2020 - Dec 2020

      Developed computer vision models based on neural network architectures Utilized U-Net and variants on semantic segmentation task of damage on company’s products Utilized Resnet and variants for feature extraction on an image classification task. The classification itself was done with both Neural Networks and KNN, Decision Trees, etc. Developed discrete event and agent based simulation models, created app for interaction and basic animation. Performed imaging techniques such as texture segmentation. Keywords: Tensorflow, ResNet, BiT, Image Classification, KNN, Decision Tree, Random Forest, UNet, UNet++, Semantic Segmentation, Simpy, OOP Show less

Community

You need to have a working account to view this content. Click here to join now