Headline

Senior Machine Learning Engineer | GenAI, LLMs & RAG | Production ML & Technical Leadership | PayPal, Walmart

About

I am a Senior Machine Learning Engineer with 8+ years of experience across PayPal, Walmart, and LatentView. I build ML systems for risk modeling, customer analytics, and enterprise AI, combining hands-on development with technical direction and team leadership. My experience spans defining business problems, designing architectures, developing and evaluating models, and leading delivery into production. At PayPal, my work includes leading a cold-start BNPL underwriting initiative for new-to-platform customers and architecting an agentic AutoML framework that orchestrates data preparation, training, evaluation, and explainability. At Walmart, I led a four-person team building DataBot, an LLM and RAG - powered platform that lets associates query enterprise data in natural language. The platform reached approximately 1,000 users, with an estimated annual productivity benefit of 17,000–20,000 associate hours. I also developed conversion and retention models to support membership growth and reduce churn. Earlier at LatentView, I built NLP classification and conversational analytics solutions to reduce manual work and improve access to business insights. I enjoy making technical trade-offs and translating experimentation into useful products Core expertise: Machine Learning • Generative AI • Large Language Models (LLM) • NLP • Retrieval-Augmented Generation (RAG) • Text-to-SQL • Model Evaluation • Python • SQL • LightGBM • XGBoost I am particularly interested in problems that combine applied ML depth, reliable production systems, and collaboration across engineering and business teams.

Location

Bengaluru

AI stack

claude code

claude

codex

Work experience

Senior ML engineer at PayPal, Nov 2025 - Present · 10 mo, Bangalore, • AutoML Agent - Multi-Agent LLM-Orchestrated AutoML| LangGraph, LLMs, LiteLLM ∗ Architected and implemented a multi-agent LLM-driven AutoML framework that autonomously orchestrates dataset ingestion, preprocessing, model training, evaluation, and explainability from natural-language requests. ∗ Impact: Reduced manual ML pipeline setup time by 70%, accelerating experimentation and enabling scalable adoption across ML workflows. • Multi-Country BNPL Launch Model| Model Strategy, Cold-Start, Risk Modeling ∗ Architected a scalable proxy-tag-based BNPL launch model to address the cold-start problem across 20 new-market launches, replacing separate per-country model rebuilds with a unified framework and defining a graduation path to transition repeat customers to history-based models. ∗ Impact: Built and validated comparative experiments demonstrating +15pp AUC improvement over the legacy model for new applicants.

Senior ML engineer at Walmart, Jul 2021 - Nov 2025 · 4 y 4 mo, Bangalore, • DataBot Framework| Text to SQL, LLM, RAG ∗ Architected and developed an LLM-powered data intelligence platform that converts natural-language queries into SQL and delivers real-time insights through a conversational interface. ∗ Designed a scalable architecture supporting domain-specific bots, Auto Suggest, interactive drill-downs, speech-to-text, and Retrieval-Augmented Generation (RAG). ∗ Evaluated and integrated LLMs including SQLCoder-15B, GPT-4o, CodeLlama-7B/13B, and Llama2-13B, along with GTE-Large, Nomic, and Gecko embedding models for retrieval and semantic understanding. ∗ Impact: Estimated $1M/year due to 17K to 20K hours of saved associate work time. • Root-Cause Mining | BERTopic, RoBERTa, LightGBM ∗ Designed an NLP-driven framework to mine tech-support logs and uncover latent drivers behind poor user experience. Applied BERTopic to cluster issue themes and used RoBERTa-based extraction to map root causes to incident outcomes. Built a LightGBM regression model to generate a 0–10 Delight Score at the ticket level, enabling real-time UX monitoring and actionable insights across issue categories and support engineers. ∗ Impact: Guided ops improvements worth $300K/year in issue resolution efficiency • Trial-to-Paid Conversion Model | Classification, Boosting ∗ Architected and implemented an end-to-end propensity modeling framework to identify trial members most likely to convert, enabling targeted customer interventions. ∗ Independently owned the ML lifecycle spanning data exploration, feature engineering, model development, ensemble modeling, experimentation, and evaluation using LightGBM, CatBoost, and XGBoost. ∗ Impact: Controlled A/B experiments demonstrated 20% uplift in conversion precision and drove $1.2M/year in incremental membership revenue. • Developed cancellation propensity Model | Boosting, Bigquery SQL ∗ Built a model to predict the cancellation likelihood of the W+ subscription of a customer in next 30 days.

Senior Data Scientist at LatentView Analytics, Jun 2018 - Jun 2021 · 3 y, Chennai, Tamil Nadu, India

Education

Bachelor of Technology, ENGINEERING, National Institute of Technology, Tiruchirappalli

11th and 12th standards., MPC, narayana junior college