Resume

Senior Machine Learning Engineer specializing in Deep Learning with 2+ years experience. Currently working at an AI focussed company ( NLP Domain ). Experienced in using Deep Learning to automate critical business processes, manage model lifecycle to ensure reliability and performance and building custom ETL pipelines.

Links/Contact Information

• LinkedIn: https://www.linkedin.com/in/rwikdutta/
• Email: rwikdutta@outlook.com
• Github: https://github.com/rwikdutta

Summary of Qualifications

• 2+ years of professional experience in Deep Learning and Machine learning working at an AI focussed company primarily in NLP domain. My primary job is to build Deep Learning models to automate business processes in production setting, manage the whole model lifecycle to ensure reliability and performance and build custom ETL pipelines according to business needs.
• Research experience in Deep Learning and Computer Vision, paper published in LNCS, Springer.
• Winner of Smart India Hackathon, a national level hackathon organized by MHRD, Govt. Of India.
• Experienced in Python, PyTorch, REST API Development using Django, Flask.

Work Experience

• Senior Machine Learning Engineer at Vaultedge Software
  • Duration: Aug ‘21 - Present
  • Responsibilities ( apart from the ones listed under Machine Learning Engineer ): <!– * Working on NLP tasks including Classification, Named Entity Recognition, Extractive Question & Answering on domain-specific Visually Rich Documents.
    • Research on improving the performance of production models and devising ways of better Quality Control on the feedback loop. –>
    • Mentoring interns and new team members. Have mentored multiple cohorts of ML interns at Vaultedge.
    • Planning and coordinating project level implementation details between team members while adding additional use-cases to VMA ( Vaultedge Mortgage Automation ) product.
    • Contributing to the architecture and design of VMA ( Vaultedge Mortgage Automation ) product.
• Machine Learning Engineer at Vaultedge Software
  • Duration: Aug ‘19 - Jun ‘21
  • Responsibilities:
    • Automate business processes in production setting using Machine Learning/Deep Learning, manage Machine Learning model lifecycle, build custom ETL pipelines according to business needs.
    • Have worked on wide range of NLP tasks including Classification, Named Entity Recognition, Extractive Question & Answering on domain-specific Visually Rich Documents. Worked on adapting state-of-the-art research methodologies, dealing with Visually Rich Documents for our use-case with great results.
    • Design internal model performance tracking solution custom to the company’s end-to-end workflows for better model performance and lifecycle management.
    • Design feedback loop systems ( integrated into our SaaS product ) for continuous feedback and improvement of our models.
  • Secondary responsibilities:
    • Design and build the Kubernetes cluster pipeline to efficiently scale ( up and down, dynamically) the company SaaS solution according to workload.

Technical Skills

• Building Machine Learning ( primarily Deep Learning ) solutions: Specific techniques that I have worked on:
  • Transformer based NLP Architectures. ( Primarily worked on NER, Classification, Extractive Q & A tasks).
  • Multi-modal Transformer Based architectures for working with Visually Rich Documents.
  • Convolutional Neural Network based architectures for Computer Vision Tasks ( Primarily worked on Classification, Object Segmentation tasks )
  • Worked with Gradient Boosted Random Forest Trees, Logistic Regression, SVM for tabular data
• Python Proficiency:. Proficiency in Python. Primary Libraries that I have worked with:
  • Numpy, Pandas (Data Analytics)
  • matplotlib, plotly, dash (data visualization)
  • scikit-learn and PyTorch (machine learning/deep learning frameworks)
  • NLTK, Spacy, Huggingface Transformers (NLP Focused Libraries )
  • fastai ( Computer Vision and NLP library )
  • Flask, Django ( Web Development; Rest API Development )
• Backend Web Development: Experience in building REST APIs which are consumed by other clients using Django and Flask.
• RDBMS and No-SQL Databases: Have worked with PostgreSQL and MySQL, MongoDB.
• Tableau (BI Tool): Have created dashboards in Tableau
• Cloud Platforms: Experience in working and deploying applications on AWS and Azure.
• Other programming languages: Experience in C, C++, Java.

Internship

• Data Science Intern at Prakshep
  • Mar ‘19 - Jun ‘19
  • Using Deep Learning to perform edge detection (Holistically Nested Edge Detection) in order to detect boundaries of agricultural plots from satellite imagery
  • Automate the process of downloading satellite imagery from Landsat 8 and analysis of given agricultural plots
  • Digitization of Identity Documents using OCR, image pre-processing and then using both logic and automation to extract the information from those cards in a structured format.

Research Experience

• Bengali Handwritten Character Classification using Transfer Learning on Deep Convolutional Neural Network.
  • Presented in 13th IHCI Conference, 2019.
  • Proceedings published in LNCS, Springer.
  • ArXiv Link.
• Review of Deep Learning Techniques for Brain Tumor Segmentation.
  • ISBN:9788182111462, Proceedings of the International Conference on Emerging Technologies for Sustainable Development, 2019 pg: 332-334

Competitions

• Winner of Smart India Hackathon, 2018 organized by Ministry of Human Resources and Development, Government of India under Department of Telecommunications ( DoT ) category
  • Worked on a solution for Department of Telecom to build a unified platform where they could identify individuals who violated DoT’s rule of maximum mobile connection limit for an individual.
• Winner of Inter college coding competition organized by my alma matter as part of Technical Fest, 2018.

Workshops Conducted

• Conducted a Code First Introduction to Machine Learning workshop remotely at my alma-mater on special invitation from my alma-mater.
  • Code link
  • Video Link

Projects

• Predicting Relevant Articles for Tweets.
  • Done as part of college final year project.
  • Show relevant articles (from “trusted” sources) for tweets. The idea is to give the social media users context for the tweets that they see so that they might not be influenced by fake news. We scrape the web for articles from trusted sources, preprocess it and when a new tweet arrives, show the most relevant news articles (from our collection of articles) for that tweet.
• Tableau Viz Dashboard - Category Wise Import and Export of Commodities by India 14-15 to 17-18.
  • Developed an intuitive dashboard using which any one can ingest India’s Export Import Scenario from F.Y. 14-15 to F.Y. 17-18 along different dimensions - Country Wise, Year Wise, Product Category Wise.
  • Link to Dashboard
• Predict Deforestation from Satellite Imagery
  • Uses data from NASA’s Landsat 8 satellite via NASA’s API to predict whether any activity related to deforestation is going on. Training data was obtained from a Kaggle competition. A Resnet-34 CNN model was used as the model.
  • The model obtained f2 score of ~0.927.
  • Link
• Backend (REST Service) for an anonymized (voice masked) audio confession app and user Q and A app
  • Developed as part of a social media app for Internal use amongst my college friends
  • Backend developed in Django (Python)
  • Link
• EDA of Codevita ( TCS Coding Competition ) 3 years results.
  • Analyzed 3 years results of TCS Codevita coding competition
  • Link
• Exploratory Data Analysis of Consumer Complaints in Brazil -Dataset: Consumer complaints for the last 5 years registered with Procon ( Consumer Protection Agency ).
  • Link
• Developed the backend (a REST Web Service) using Django for a mini social networking site that we had made for Teacher’s Day (Link)
  • Link

Online Courses

• fast.ai by Jeremy Howard
• Coursera: DeepLearning.ai by Andrew Ng
• Coursera Certification: Machine Learning by Andrew Ng (Stanford University)
• NPTEL Certification: Introduction to Machine Learning (IIT-Kharagpur)
• NPTEL Certification: Introduction to Data Analytics (IIT-Madras)

Formal Education

• University: Maulana Abul Kalam Azad University Of Technology
  • Degree: Bachelor of Technology in Computer Science and Engineering
  • Duration: 2015-2019
  • DGPA: 8.28/10
• School: Delhi Public School Siliguri
  • Board: CBSE
  • 12th Percentage: 85.6%
  • Duration: 2013-2015
• School: Don Bosco School Siliguri
  • Board: ICSE
  • 10th Percentage: 93.6%
  • Duration: 2003-2013