Please enable JavaScript.

Coggle requires JavaScript to display documents.

Myself - Coggle Diagram

- - - - 2D CAD
        
        Cloud
        
        Server
        
        Nodejs/Python/C++
        
        Outscale
        
        ODA C++ SDK microservice for kernel
        
        Create a json based scene graph of all grpahic elements
        
        Authentication (CAS-Central auth server)
        
        Database/Storage
        
        mysql - tables-user
        
        PLM for storing objects
        
        Client
        
        In house JS library which understands our scenegraph. webgl
        
        Angular JS started with V7 in 2018 and now using v10 since 2020
        
        Desktop
        
        ODA SDK
        
        Qt 5.12 lts
        
        OpenCV and AI - vectorization
      - ERP
        
        Cloud
        
        Client
        
        Unigui
        
        Server
        
        IIS with Delphi
        
        Desktop
        
        AI based PO matching. Using AI to detection PO, Invoice number and other details from vendor invoice and do a 3 way matching with PO, Invoice and goods receipt note (GRN). Object detection instead of OCR to recognize tables and fields similar to object recognition in images and then extract the text out of it using AI as well
    - - ERP - 2 teams
        
        Desktop - 2 teams
        
        Cloud 2 teams
      - CAD - 5 teams
        
        Desktop - 3 teams
        
        3-4 devs and QA s
        
        Cloud - 2 teams
        
        3-4 devs and QA s
    - - SSE
      - Team lead
      - Manager
      - Senior manager
      - To be Director
- - - - Benefits
        
        Fresh look at things
        
        See all options which we did not think
        of earlier/ Different perspextive
        
        Solutions pop out
      - Examples
        
        Vectorization. we had a lot of issues with brute force
        vector generation from bitmaps and decided to see
        if we could used AI which requires a lot of painstaking trials
        and fails.
        
        Licensing: we had issues with our internal
        implementation and decided using Flex licensing
        
        Spatial: In the middle of a BIM implementation we realised
        that we are reinventing a lot of things that could be
        done using spatial SDK
    - - Examples
        
        Vectorization. Different AI stuff failed
        
        Lot of protoyping
        
        Authetication
        
        Collaboration APIs
        
        Different designs
  - - - Technical conflict
        
        Examples
        
        Mongodb or sqllite for user collab info in DraftSight
        
        Spatial SDK or ODA SDK for BIM
        
        Which technique to use for object detection(YOLOV3) and semantic segmentation (R-CNN, VGG-16).
        
        Resolved by prototyping different ideas and seeing
        performance