About

I lead applied AI and analytics at Walmart and am pursuing an MS in Artificial Intelligence at Georgia Tech. My research studies how to evaluate and trust AI agents, and how to draw sound conclusions from imperfect data.

Raised in Bengal among its painters and poets, I believe a good model, like a good poem, should say exactly what it means.

  • Deep Learning
  • Natural Language Processing
  • Explainable AI
  • Agentic AI
  • Decision Science
  • Statistics

Publications

2027
  • SG Chowdhury. Beyond the Illusion of Power: Calibrating Quasi-Experiments in Observational IS. 60th Hawaii International Conference on System Sciences (HICSS-60).Jan 2027
2026
  • SG Chowdhury, P Maji. Risk-Aware Adaptive Evaluation: Finding High-Impact Failures Under Limited Budgets. NeurIPS 2026 Workshop on Evaluation of Interactive Agents.Dec 2026
  • SG Chowdhury. Closing the Loop on Contrail Avoidance with Satellite Verification. NeurIPS 2026 Workshop on Tackling Climate Change with Machine Learning.Dec 2026
  • SG Chowdhury. Commitment Hierarchies under Intent Revision: A Belief-Revision Account of Salvage in Tool-Use Agents. 27th International Conference on Principles and Practice of Multi-Agent Systems (PRIMA).Dec 2026
  • SG Chowdhury. Agentic Share-of-Search: A Multi-Agent AI System for Competitive Decision-Making in LLM-Mediated E-Commerce. 57th Annual Conference of the Decision Sciences Institute (DSI).Nov 2026
  • SG Chowdhury. Selection Bias Correction in Retail Intelligence. 15th International Conference on Data Science, Technology and Applications (DATA), Porto.Jul 2026
2014
  • SG Chowdhury, S Routh, S Chakrabarti. News Analytics and Sentiment Analysis to Predict Stock Price Trends. International Journal of Computer Science and Information Technologies 5(3): 3595–3604.

Patents

  • RL-Based Framework for Refresh Frequency Determination
    U.S. patent applicationFiled, pending
  • GenAI-Powered Private Brand Matching
    U.S. patent applicationFiled, pending

Experience

Applied AI scientist and manager with over a decade of experience leading decision science and full-stack analytics, with a focus on econometric methods (IV, DiD, quasi-experiments), experimental design and production LLM systems.

Walmart Inc.Jan 2019 – Present
Principal Data Analyst / Senior Manager, Advanced Analytics
  • Defined and executed the multi-year data science roadmap for competitive market intelligence (pricing, shipping, assortment, adTech), aligned with eCommerce P&L goals and VP+ stakeholders across Product, Engineering and Merchandising.
  • Architected an LLM-based product matching engine that extracts Private Brand entities from multimodal catalog data, combined with IV regression to improve competitive price positioning in low-elasticity categories.
  • Owned the end-to-end model lifecycle for Gemini Pro, Flash and Gemma (training pipelines, prompt optimization, knowledge distillation, deployment and monitoring) on PySpark infrastructure serving tens of millions of items daily.
  • Designed a data freshness framework using novel adversarial bandit algorithms (Exp3-Acc/AccIX) for large-scale knapsack optimization under budget constraints.
  • Applied causal inference (IV, DiD, RDD, geo-randomized cluster trials) to pricing and promotion strategy, lifting seller GMV and cutting crawl costs while maintaining high price freshness on key value items.
  • Built a real-time deal intelligence system and established team standards for model validation, experiment design and responsible AI, including human-in-the-loop review for high-GMV decisions.
Ernst & YoungDec 2016 – Jan 2019
Consultant, Advisory – Data & Analytics
  • Built analytics products with C-level executives for finance, audit and tax professionals, substantially cutting project turnaround time.
  • Designed A/B tests and controlled experiments for marketing campaigns and product upgrades; led SQL, SSIS, Alteryx and Python boot camps across business units.
Seven Lakes TechnologiesDec 2015 – Dec 2016
Senior Software Engineer, Database
  • Designed modules for two flagship products for the Oil & Gas sector and managed new customer accounts, contributing to revenue growth.
MAQ Software2014 – Dec 2015
Indian Institute of Management Calcutta2013 – 2014
Research Intern
  • Optimized portfolio allocation with a multi-objective genetic algorithm (C++, Pareto front) using a Sharpe-ratio fitness function, outperforming the state of the art on risk-adjusted returns.

Memberships

  • IEEESenior Member
  • Association for Information SystemsMember
  • Decision Sciences InstituteMember

Education & More

Education

Skills

  • Python (PyTorch, scikit-learn), PySpark, SQL
  • LLMs, RAG, agentic AI, knowledge distillation
  • Causal inference, A/B testing, time series
  • Deep learning: CNNs, Transformers, VAEs, GANs, diffusion

Awards

  • Walmart: 6 Excellence Awards, 2 Bravo Awards, Employee of the Month
  • EY Exceptional Client Services Award

Certifications

  • NVIDIA: Rapid Application Development with LLMs
  • NVIDIA: Building LLM Applications with Prompt Engineering
  • NVIDIA: Building RAG Agents with LLMs
  • Building Agentic AI Applications with LLMs

Contact

I am always glad to hear about research collaborations, speaking invitations, or a good book recommendation. Write to me at spandan1305@gmail.com.