Dipanjan Nag

Developer, Artificial Intelligence
Noida Area, Xavient Digital- Powered by Telus International
  • Automatic speech recognition pipeline with Kaldi, DeepSpeech and other Deep learning approach.
  • Analytics for Digital Learning Product. Spark backend on AWS stack.
  • Initially end-to-end DevOps on Jenkins stack for Fortune 500 Healthcare Client.
  • Pre-Sales for Healthcare Domain on IoT connected device and anomaly detection on Sensor Data.
  • Machine Learning with Python. Deep Learning. Keras (Tensorflow/Theano)
  • Language modelling
  • PySpark
  • Still good at c++

Machine Learning Engineer | Xavient Digital- Powered by Telus International | Mar, 18 – Now
In house Speech Recognition engine at scale. Language modelling. Deep Learning based speech recognition (Deepspeech etc.)

Developer | TCS | Sep,16 – Feb, 18
Getting raw user interaction data from AWS s3 via spark. Extracting actionable insight for a digital learning product. Passing through Machine Learning Pipeline to recommend courses to users in line with unit and project requirement.

Developer | TCS | Mar, 16 – Sept, 16
Implemented end-to-end Devops implementation for a Fortune 50 Healthcare client. Efficient in Shell Scripting.

Developer | TCS | Dec, 15 – Mar, 16
Pre-Sales activity for various healthcare client. PoC on Azure ML, Apache Camel, Spark, Azure Stream Analytics.

Work outside Regular Job (mostly in college):
  • Agro Data Agility: Gives an intuitive UI like Azure ML Studio to create powerful dataset from the USDA dataset. Providing tailored USDA data as Data As A Service. All type of queries can be run on resulting datasets. @GitHub


  • Summary:It understands entities, concepts and keywords. Summary For HipChat actually brings Natural Language understanding in the scene to provide sensible summary of current room. Don’t feel alienated after a break. Store [DISCONTINUED] Open Sourced


  • AtroSeer: AtroSeer tracks news, tweet and other relevant data in real time and matches pattern from past data (available via USAID) to find out where the next atrocity will occur. Details is available on My Website. This project was finalist for Microsoft Imagine Cup.


  • eiE: Location aware social app for windows phone. To tag remarks to a location. Detail is available here. Download it from store.


  • Hasher:Hasher is a very tiny nuget package to generate Guid salt and corresponding password hash (sha2) and later to verify if the password is correct w.r.t. stored salt and hash value. Available as Nuget package.


  • libann:A library to implement and train neural network using c++ amp. It uses GPU with a++ amp for calculating result in each layer. For brunched network it creates separate threads via std::thread. The code is available at Github


  • B. Tech in Computer Science and Engineering
  • The most ignored: May be the most powerful, effective at National Conference on Computing and Cloud Security – 2012. available Here
  • Fifa
  • SciFi/Fantasy book (Favorite Red Rising Triology)


If you prefer a pdf version you may download it from here.

1 comment for “Dipanjan Nag

  1. Sai Prasanth Jeevalingam
    May 6, 2019 at 6:05 am

    Hello Dipanjan,

    I came to know about you from Kaggle.

    I am Sai Prasanth Jeevalingam from Bengaluru, I have 9 years of IT experience in the field of Infrastructure Management domain – STORAGE Administrator.

    I am a non-programmer, but i am willing to take Data scientist filed on my interest.

    I am in need to your guidance. Can you please guide me what will be starting point i need to focus on to become entry level Data scientist.

Leave a Reply

Your email address will not be published.