Feb 2023 – Oct 2025 · 2 yrs 8 mos

• ■ Project 3: Carepack Forecasting for WorldWide HP (Approach - Generation Over Generation Forecasting Technique, Machine Learning & Python) (Ongoing) [Received Team Hero Award]
• Spearheaded Carepack Forecasting Model, enhancing budget planning accuracy by 12%, improving service cost efficiency.
• Designed GoG Forecasting Dashboard using Python & PowerBI, reducing manual work load by 40%.
• Carepack budget is decided based on the forecasting of Service Install Base (SIB) and Annualize Intervention Rate (AIR). Accurate SIB forecasting is one of the prime requirements to understand the future AIR. According to the forecasted value of AIR upcoming budget is decided.
• ■ Project 2: Warranty & Carepack GoG Dashboard (Approach: Statistics, Python & PowerBI) [Received Official Recognition]
• Designed GoG Actuals Dashboard using Python & PowerBI, reducing manual work load by 95%.
• An automated method of displaying Generation-Over-Generation (GoG) charts for Install-Base, Interventions, Annualize Intervention Rate (AIR), and other correlated metrics. First processing raw calendar view data into GoG format using the python programming based on the first month of Fiscal Year as “November”. Next, PowerBI play a role to visualize the charts.
• ■ Project 1: Support Net Promoter Score (sNPS) Dashboard (Technology - Python and PowerBI)
• Developed Support NPS Dashboard (NLP & Statistics), enabling leadership to make data-driven customer experience improvements.
• Features: Generation Over Generation sNPS View, Promoter vs Detractor Pie Chart, Year over Year sNPS% view, Sentiment Scores, PL-level sNPS% view, Worldcloude on the Detractors' sNPS Comments, Trun Around Time (TAT) vs sNPS Scores Correlation.

May 2021 – Feb 2023 · 1 yr 9 mos

• ■ Project 5: Contact Center Control Pannel: Lead metrics analysis and agent Performance Analysis (Approach: NLP & Stat)
• Led Contact Center Analytics, optimizing agent efficiency and reducing operational inefficiencies via AI-driven insights.
• Automatic way to find the scope of world-wide waste reduction using several correlations of different lead metrics such as NMU, PRE, Live Lense, PPMC and PPMD. “NMU” -> No Material Used.
• ■ Project 4: NMU India Drilldown Menu Insight Dashboard (Approach: Pattern Finding) – (Received BOMBERO FIRE-FIGHTER Award & Innovation Award)
• Pioneered India NMU Drilldown Dashboard (NLP \& RPA), automating issue categorization, reducing manual analysis time by 70%.
• For India-NMU cases, we did rigorous analysis in terms of finding "pattern" to from work-order-resolution note to segregate “Primary” and “Secondary” Issues (all issues considering).
• ■ Project 3: India Field NMU (Approach: NLP & RPA) - (Received Director-Level Recognition)
• When a case is created by agent and transaction happen in that case without using any spare-parts then its known as NMU. For India-NMU cases, we did rigorous analysis in terms of finding "# of interventions", "insights on gray issues like No-Boot, No-Display, No-Power", and so on.
• ■ Project 2: Advisory Adherence (Approach: Text Mining & RPA)
• “Advisory” or “Wise” code tells you a proper diagnostic step for a specific problem. Agent or field engineer needs to put the advisory code before closing a case. Here, we check the number of misses of advisory code for each issue using text mining approach.
• ■ Project 1: Einstein-NMU Categorization (Approach: Machine Learning)
• To investigate each case, need to identify issues’ department. For example, whether the issue is happened by either “Contact Center” or “Field” or “Supply Chain " so on and so forth. To solve the problem, a machine learning based approach has been developed with 85% accuracy.

Mar 2019 – Apr 2021 · 2 yrs 1 mo

• ■ Project 6: Social Listening (Received Official Recognition)
• A end-to-end system is developed which contains from a scrapping BOT (data from HP Forum) to a insights findings tool of public data on HP products. This system will help the organization by hitting the particular "Waste" area which will improve the quality of the future products.
• ■ Project 5: Predictive Analytics (Flash Intervention during disaster time frame)
• A two-layer predictive model is proposed to develop the complete system which can Flash the Interventions during disaster time frame like pandemic & lock-down situation. To achieve the model we use ensemble of “regression model with causal effects”.
• ■ Project 4: QPS (Quality Predictive System) (Received Official Recognition)
• An alert system is developed to alert managers/operations-managers on the current spike of dispatches in comparison with history using Six-Sigma SPC (Statistical Process Control) methods. We are also developing an IDP (Intelligent Dispatch Process) by finding accurate issues from the conversation of agent and customer using Deep Neural Net.
• ■ Project_3: Eagle (Automated quality analysis of consumer review)
• Proactive product and service quality analysis using "sentiment analysis" & "topic modelling" on consumer review. In this project, we are also trying to understand the "what customer wants?" (customers' expectation) which provides meaningful insights to improve the quality of product and service.
• ■ Project_2: Early warnings Analysis
• Analysis of product's quality before launching or just in 1st week of launching.
• ■ Project_1: Automation Work (Received Official Recognition)
• To reduce manual efforts and save the cost of organization, automation is highly desirable. Example of some tools are: automatic review scrapping from e-commerce sites, automatic recommended spare list (RSL) validation, and so on.

Apr 2018 – Mar 2019 · 11 mos · Bangaluru, India

• ■ Project: Reconciliation Management System (RMS) for HSBC & OCBC Banks
• RMS provides a matching solution between Source and Ledger transactions which is a major problem for two banks such as OCBC & HSBC.
• To solve the above problem, we are using some machine-learning, feature-engineering and rule-based techniques which are as follows:
• Machine Learning Algorithms: Decision Tree, Linear Regression, Random Forest.
• Feature Identification Techniques: Bag-of-words, N-Grams, making features based on business logic.
• Rule-Based Techniques: Based on our requirement, modifying an existing algorithm such as "Subset-Sum" and using other few existing algorithms like "Longest Common Subsequence (LCS)", "Longest Common Substring (LCS)", and so on. Finding out many new patterns using Regular Expression.

Apr 2016 – Mar 2018 · 1 yr 11 mos

• ■ Project: IIT Kharagpur > E-Business Centre of Excellence > Data Analytics > Sentiment Analysis of UGC
• My Research and Development Objectives :
• 1. Developed a procedure for automatic creation of n-gram sentiment dictionary.
• 2. Developed a procedure for cross-domain sentiment analysis using the above.
• 3. Developed a support system using the above dictionary and procedure for aiding the buyer’s purchase decision.
• Keywords: User-Generated Content (UGC), Lexicon-based Sentiment Analysis, Aspect Based Sentiment Analysis, NLP, ML, Statistics.
• More Details Available at: http://www.iem.iitkgp.ernet.in/eco/index.html

Jun 2015 – Mar 2016 · 9 mos

• I was a researcher of "HCI-BCI" (Human Computer Interaction - Brain Computer Interface) Lab.
• I was involved in two projects which are as follows:
• 1) Virtual Keyboard Standardization of Indian Languages for smart phone (SVL): The main work was in this project was to build an XML file using some "easy", "medium" and "hard" words and sentences to test and standardize some existing keyboards for mobile. To test and identify the best and handy keyboard, we have done some survey by visiting several schools, colleges including our institute and by involving students of those institutes. Finally, to decide the best one, we calculate and prepare a statistical measure by using some parameters from the users typing such as error rate, number of back space etc.
• Note: We worked on two Indian Languages: a) Santali b) Oriya.
• 2) Brain Computer Interface: In this project, our main objective was to detect left-hand and right-hand movement from the "electroencephalography" (EEG) single which is generated from our thinking. To collect the data we used an EEG collector device (EMOTIV EPOC+ - 14 Channel Wireless EEG Headset).
• Based on this work, we have published a conference paper which is attached below.

Sep 2014 – Jun 2015 · 9 mos · Hyderabad

• I was a developer of a project namely as "XL-CATLIN".
• Project: XL-CATLIN is a big insurance company in USA. Each insurance company maintains three centers such as "Policy Center", "Claim Center" and "Billing Center". To handle each centre automatically by user-interface, needed to develop some web application.
• To solve the above problem, we took initiative to develop such kind of system using "Guideware" software package because this software has some inbuilt functions for all three centers. That's why Guideware was the best choice for the project.
• I was fully involved with "Policy Center" which most vital part among three sections in the project.