Hareesh G

• Responsible for driving cross functional data science initiatives towards reducing enterprise wide past due receivables.
• Identification of anomalous purchase order patterns: LCS based sequence mining algorithm was developed to upfront identify anomalous purchase order patterns.
• The anomalous purchase orders identified by algorithm had a significant impact on downstream collection activity.
• Annual opportunity was estimated to be $10Mn in working capital & $5Mn in bad debt provisioning.
• R- Shiny based prototype application was developed to facilitate IT implementation.
• Identification of order to cash friction points: White box models developed at invoice level has uncovered friction points in order to cash cycle like payment terms mismatch, tax related issues etc.
• The identified friction areas were further drilled down collaboratively with operations team to root cause and fix the issues.
• Friction areas were then tracked through weekly reports driving action from operations team.
• The initiative has been successful in reducing $70Mn in past due receivables from its inception on the identified friction areas.
• Preventing duplicate creation of customer accounts : Leveraging fuzzy matching techniques to identify & prevent duplicate customer accounts from being created.
• Payment Leakage detection : Developed & operationalised statistical solution to proactively detect & review anomalous payment activities.
• Senior Data scientist, (Vmware part of Dell Technologies).
• Setting up DS CoE practice, execution & engagement frameworks, infrastructure.
• Responsible for ideation , stakeholder alignment and execution of ML use cases in IT org.
• Developed NLP based case classification and automated action algorithm resulting in ~30% improvement for quoting team’s productivity.
• Developed an automated PDF content recognition solution for automating quality checks on contracts using tesseract,opencv, magick and natural language patterns improving operational efficiency by 50%.
• Auto matching algorithm to harmonize customer accounts from different source systems achieving ~90% accuracy with custom built algorithm.
• Developed transfer learning based sentiment analysis model to track sentiment on deals using forecast comments (XLNet & BERT based on PyTorch framework).
• Proactively identified ~20Mn per Qtr worth opportunities that are at risk of non-renewals, by developing hybrid approach based on rules and ensemble modelling(random forest+gradient boosting+ridge & logisitic classifier).
• Developed “Industry to Focus” matrix to enable sales teams for targeting specific accounts based on YoY numbers & other KPIs along with trend/seasonality guided forecasts(~4% MAPE).
• Conceptualised custom propensity scoring model for targeting customers having higher potential to do migration to new networking product version for channelling marketing and support activities.
• SPADE analysis for identifying frequent sequential product purchase patterns for customers by Geo/FY etc.
• Docker based deployments of ML projects as service using Plumber/ Flask APIs.
• Mentoring colleagues through the DS journey on execution and engagements.

Order to Cash Analytics

Responsible for driving cross functional data science initiatives towards reducing enterprise wide past

due receivables

• Identification of anomalous purchase order patterns: LCS based sequence mining algorithm was

developed to upfront identify anomalous purchase order patterns. The anomalous purchase orders

identified by algorithm had a significant impact on downstream collection activity. Annual

opportunity was estimated to be $10Mn in working capital & $5Mn in bad debt provisioning. RShiny

based prototype application was developed to facilitate IT implementation.

• Identification of order to cash friction points: White box models developed at invoice level has

uncovered friction points in order to cash cycle like payment terms mismatch, tax related issues etc.

The identified friction areas were further drilled down collaboratively with operations team to root

cause and fix the issues. Friction areas were then tracked through weekly reports driving action from

operations team. The initiative has been successful in reducing $70Mn in past due receivables from

its inception on the identified friction areas.

• Preventing duplicate creation of customer accounts : Leveraging fuzzy matching techniques to

identify & prevent duplicate customer accounts from being created

• Payment Leakage detection : Developed & operationalised statistical solution to proactively detect

& review anomalous payment activities.

Performance tracking of JDA Fulfillment system

• Developed multiple rule based models in SAS to root cause the drivers of inventory difference in JDA system with respect to legacy

system.

• Developed exception reporting in SAS to alert the business users on a daily basis on the possible deviations in order-shipment

recommendations in JDA Fulfillment system.

• Automated data extraction from various sources like Hadoop, DB, Netezza and oracle databases, developed optimized pass through

queries thereby saving approximately 5 Hrs. in data extraction

Key Responsibilities & Initiatives (Cognizant Analytics)

Other Projects in Analytics

Call Center Transformation

• Nurturing a team of six analytics resources towards delivering end to end customer analytics solutions for a call center transformation

project

• Key Focus Areas: Segmentation, Cross Sell Analytics, Acquisition & Retention Modelling

• Developing statistical segmentation to identify homogenous customer groups, conceptualizing and developing a personalized product

recommendation system, developing an acquisition model leveraging advanced data preparation techniques like fuzzy matching,

identifying indicators of customer churn.

Trade Promotion Optimization

• Conceptualizing and implementation of analytics based performance tracking of trade promotions for key product categories for a C&G

customer

• Baseline sales computation for key products categories taking into account trend and seasonal effects

• Tableau based interactive story board has been developed to see promotion performance at product/distributor/route levels

Operational metrics’ impact on customer satisfaction

• Analytics techniques like hypothesis testing, correlation & regression analysis were used to understand the impact of an airport’s

operational performance on its customers’ satisfaction level

• Collaborated with an Indian vendor in developing alternative to an imported electronic module leading to cost savings of INR 50Lakhs.

• Developed software using Visual Basic for automating the tuning procedure of Power amplifiers and thereby reduced the cycle time by

4 times.

• Improved the shop floor production rate by 13% by streamlining the Bill of Materials, nomenclatures, drawings and monitoring the

critical, long lead time items.

• Implemented ERP system (IFS) in shop floor to streamline the production activities like shop order generation and closing, rejection of

faulty materials, generation of inspection memo, finished goods report etc.

• Functioned as a single point of contact for resolving supply issues with international vendors.

• Conducted training programs and roadshows in various customer locations as a part of revitalizing and repositioning the product brand.

• Undertook indigenous upgradations for the transmitter to suit the new customer requirements which resulted in bulk order placement

worth INR 300 Cr from Indian Navy.