Kumar Saurav
AI/ML Leader | Staff Data Scientist @ Walmart Global Tech | GenAI, LLMs, Agents, Search, Ranking & Recommender Systems | Driving AI Strategy and Business Impact
Greater Bengaluru Area
18+ years of work experience
Sep 20215 yrs 1 moPresent
Bangalore Urban, Karnataka, India
Foundational Language Model:
o Designed and implemented foundational LLMs with pre-training and fine-tuning strategies tailored for Walmart’s catalog data.
o Enabled high impact use cases, including product attribute extraction and duplicate item detection in catalog.
Marketplace Seller Risk:
o Listing Fraud Detection: Conceptualized and developed a system to identify and reduce counterfeit item listings caused by branded content misuse.
o Utilized Text and Image modalities to detect counterfeit items in catalog
Marketplace Assortment Growth – Recommender System
o Developed and launched a two-tower deep learning recommender model to drive assortment growth in Walmart.
o Delivered a single solution addressing assortment recommendations for existing items and seller recommendations for new products (not a part of Walmart).
o Filed for Solo Patent with Patent Reference Number: 8168US01
o Chosen as one of the research papers (out of 678) in Walmart SparkTech’23 Summit, an internal technical paper competition, for in-person presentation, hosted in Bentonville, US
Customer Intent Classification
o Launched BERT-based models to classify customer intent (pre/post-transaction) and improve seller experience.
Seller Empowerment via GMV prediction
o Conceptualized and developed solutions to predict future GMV for recommended items
o Helped seller make an informed decision about acceptance/rejection of recommendations.
Others:
Multiple predictive models using ML to solve different business problems.
Oct 20188 yrsPresent
Bengaluru Area, India
Oct 20162 yrs 1 moOct 2018
Bengaluru, Karnataka, India
1. Worked on problems related to cross-selling, IVR call classification, recommendations.
2. One of the few members to introduce Deep learning for the projects within organization.
3. Worked on Multi-label classification, Bayesian Optimization etc.
Cross-Sell Initiatives
o Developed classification models for APAC/EMEA countries using multiple ML techniques.
o Delivered multiple Cross-sell targeting models for different lending products being offered by Citi for Russia/North America to optimize the offer pallet.
Multilabel Classification Models – Call Intent Categorization
o Deployed text mining techniques to classify customer calls into multiple categories using binary relevance and chaining methods.
Balance Growth Model
o Developed a Logistic Model to identify the customers who are more likely to grow their balances in future with Citi
o It has been implemented to identify Balance Growers and proactive RM (Relationship Manager) assignment and Optimization.
o Estimated yearly Revenue impact of $3MM.
Balance Attrition Model
o Developed attrition Model to trim down losses due to balance eroding households by proactive targeting of attrition prone households in future.
o Model has been implemented with estimated yearly Revenue impact of $5MM.
Jan 20142 yrs 10 mosOct 2016
Bangalore,India
Jul 20121 yr 6 mosDec 2013
Bangalore
Mar 20111 yr 6 mosAug 2012
Bangalore
Worked mostly on Exploratory Analysis, Modeling and Segmentation
Oct 20082 yrs 6 mosMar 2011
1. Brand awareness and market segmentation study
2. Brand architecture and brand positioning analysis
3. Exploratory data analysis
May 20063 mosJul 2006
Summer internship at Ohio State University
Kumar’s AI Stack
Education
B.Tech and M.Tech, Agricultural Engineering
Indian Institute of Technology, Kharagpur
20035 yrs2008
Chinmaya Vidyalaya, Bokaro Steel City
20002 yrs2002
Join Kumar and 10k+ other professionals on
Tal is the app where bosses meet engineers & builders directly, without middlemen
Download Tal Boss