Principal, Data Scientist

Overview

Principal Data Scientist on Walmart's Data Foundations and Avoid team in Customer Care Tech, based in Bengaluru, India. The role defines problem statements and solution direction, owns the full lifecycle from identifying high-leverage bets through production rollout, and builds an AI-powered platform that turns customer and operational signals into actionable insights.

What you'll do

• Identify problems, patterns and root causes behind customer conversations.
• Build alignment across product, engineering, business and data science teams in the US and India, and change long-held assumptions using evidence.
• Turn ambiguous business goals into precise objectives, model architectures, loss functions, and offline and online evaluation frameworks.
• Lead data engineering over billions of interactions, large-scale distributed GPU training, and inference optimization (distillation, quantization, caching, custom kernels).
• Take models through A/B tests into production, measure business impact, and turn wins into platforms reused by many teams.
• Mentor scientists and engineers, lead design and research reviews, and represent Walmart through publications, patents, and open-source work.
• Build agent evaluation frameworks: scalable LLM-as-judge and model-based graders calibrated against human judgment, retail-specific benchmarks, simulated users and environments for multi-turn and tool-using agents, and automated red-teaming.
• Build frameworks, standards, and tools that raise the bar for code quality, reliability, and operational excellence.
• Lead technical explorations, evaluate emerging technologies, and incubate solutions.
• Partner with product, UX, program management, and global engineering teams.
• Embed security, performance, and fault-tolerance into every layer of the stack.
• Own and deliver solutions that move critical business metrics.

What you'll need

• Bachelor's, Master's, or PhD in Computer Science, Machine Learning, Statistics, Mathematics, or a related field.
• 10+ years of experience applying cutting-edge deep learning, with deep expertise in at least two of: Recommender Systems, Language Models, Multimodal Models, Deep Reinforcement Learning.
• Hands-on experience with distributed GPU training, including data, tensor, and pipeline parallelism, FSDP/DeepSpeed/Megatron-style frameworks, and mixed precision.
• Experience with low-latency inference serving (e.g. vLLM, TensorRT-LLM, Triton Inference Server) and model compression.
• Able to profile and debug performance across the stack, and to optimize kernels in CUDA and/or Triton.
• A track record of taking deep learning models from idea to production at scale, with measurable business impact.
• Expert in Python and PyTorch (or JAX), and comfortable with large-scale data processing (Spark, BigQuery, or similar).
• Proven ability to lead through influence: setting technical direction and driving adoption across teams without direct authority.
• Minimum qualifications (any one option): Option 1: Bachelor's degree in Statistics, Economics, Analytics, Mathematics, Computer Science, Information Technology or related field and 5 years' experience in an analytics related field. Option 2: Master's degree in Statistics, Economics, Analytics, Mathematics, Computer Science, Information Technology or related field and 3 years' experience in an analytics related field. Option 3: 7 years' experience in an analytics or related field.

Details

• Location: Bangalore office (Block-1, Prestige Tech Pacific Park, Kadubeesanahalli Village, Varthur Hobli, India) for daily work, with flexibility to manage personal lives.