Summary
Overview
Work History
Education
Skills
Patents & Research Papers
Timeline
Generic

Ajay Chatterjee

Benglauru

Summary

Data Scientist & Software Engineer with 10+ years of experience specializing in solving complex, ambiguous problems. I excel in leading research initiatives in Machine Learning, from identifying critical issues and collecting data to developing and experimenting with cutting-edge methods. Proven track record in translating research prototypes into fully functional products. Currently as a data scientist, I lead various Sustainability efforts and Cloud Hardware deployment automation at Microsoft.

Overview

10
10
years of professional experience

Work History

Data Scientist

Microsoft
Bengaluru
06.2021 - Current
  • Created three sustainable products - Azure Emission Impact Dashboard, M365 Emission Impact Dashboard, and Azure Spend Tracker. These products aim to provide detailed insight into emission impact based on service utilization, including Scope 1, Scope 2, and Scope 3 emissions.
  • Designed and implemented multiple assessment methodologies to accurately measure Microsoft's Scope 3 carbon emissions across the supply chain.
  • Developed a novel graph-based machine learning model to identify similar hardware components by leveraging hardware structural data. This method helped to estimate the emission of 60% of Microsoft's cloud hardware component, allowing emission profiling across 7K+ hardware products. (Patent Grant Pending, in-process for publication)
  • Calculating emission profile for a single hardware takes 2-3 days of an SME effort due to complex hardware structure used in Microsoft Datacenters, scaling that for 7k+ hardware configuration is a time consuming task. To alleviate the problem, I developed a flexible tree machine algorithm that helped to identify 125 unique hardware with confidence for emission profiling and use them as a proxy for other configurations. This approach provided a 50x gain, saving time and cost.
  • Currently leading optimization effort of IFS (Intelligent Fulfillment System) within Microsoft Cloud Supply chain organization. I am responsible for managing the mixed-integer optimization system, translating business requirements into mathematical constraints, objectives, and defining metrics to measure the improvement of business objectives over time.

Senior Machine Learning Engineer

Legato, Part of Anthem Healthcare
Bengaluru
06.2020 - 06.2021
  • Developed conversational AI interface for newborn parents to create new business opportunities in the health insurance space. Within two months of deployment, 50K members started using the interface, and 60% of conversations are successfully handled monthly.
  • Developed a new few-shot learning model for user intent classification. The proposed method provided 30% better accuracy than available market products.
  • Applied advanced analytics to detect the risk of heart attack of members based on health claim data. Proposed approach is able to save yearly $100M Million+ at least with reduced number of angioplasty and bypass procedure claims. Forecasted result used to overreach members through digital channel to better manage health.
  • Developed recommendation system for one complex dashboard application. The proposed system helped to identify users' journey across various pages of the dashboard and load them in the background to increase customer satisfaction. Through customer click-rate metrics, 65% of time pages are auto-loaded, saving 5-10 minutes of time page load wait time.

Research and Development Specialist

Accenture Lab
Bengaluru
09.2014 - 06.2020
  • Developed Natural language interface to query large domain knowledge graph for end user consumption. Early industrial application in 2016 for VCs and senior VPs to understand shift of technology domain with natural language questions.
  • Developed innovative product, Diagnostic Virtual Agent, in Conversational AI space. 3 patents are granted and one research paper accepted in International conference in Computational Linguistics. More than 5+ large clients using product for chat bot knowledge management and increase conversation
    success rate.
  • Deployed patented text based classification pipeline as part of intelligent email routing for one of leading US bank. Proposed approach provided 25% more accurate results than commercial products
  • Developed large lease processing application to automatically extract information from 100+ pages documents. System reduced more than 30 hours of man efforts, per lease document.
  • Developed research prototype for video to text summarization for sensitive (NSFW) contents. Solution developed as part of social good effort and reduce mental workload for people working in content moderation space. Received global award for innovation business case in social transformation

Education

Bachelor of Engineering - Compute Science

IIEST, Shibpur
Kolkata
08-2010

Skills

  • Natural Language Processing
  • Machine Learning
  • Deep Learning
  • Graph Machine Learning
  • Mixed Integer Optimization
  • Data Visualization
  • Metrics Definition
  • Supply Chain

Patents & Research Papers

  • Computing System for predicting substitute component of assembled product. (Filled, July 2023)
  • System and method for generation of conversation graphs. (Granted)
  • System and method for training a virtual agent to identify a user's intent from a conversation. (Granted)
  • System and Method for interactively guiding users through a Procedure. (Granted)
  • Automated and optimal encoding of text data features for Machine Learning Models. (Granted)
  • Intent Mining from past conversations for Conversational Agent (COLING, 2021)

Timeline

Data Scientist

Microsoft
06.2021 - Current

Senior Machine Learning Engineer

Legato, Part of Anthem Healthcare
06.2020 - 06.2021

Research and Development Specialist

Accenture Lab
09.2014 - 06.2020

Bachelor of Engineering - Compute Science

IIEST, Shibpur

Similar Profiles

CREATE PROFILE
Ajay Chatterjee