Headline

Senior Director/Commerce Platform AI Research Leader, ACM Eminent Speaker India 2025

About

AI R&D Leader, building next-generation AI based agentic/decisioning systems with post-transformer architectures – taking research ideas to production. ACM Eminent Speaker 2025 (https://india.acm.org/education/learning/eminent-speaker-program) 1. creating IP - patents and publications in top conferences including NeurIPS and AAAI. 2. contributing into Microsoft products - some of our foundational vision models shall be incorporated into Azure ML Foundry. 3. Solving hard problems - working with top cloud enterprise customers of Microsoft to solve real business problems of cloud customers in areas of computer vision, anomaly detection, recommenders etc. 4. Responsible AI - am the Lead Responsible AI champion for C+E and have reviewed more than 50 ML models to ensure transparency, accountability, privacy etc. Worked with Gen AI teams to ensure enough red team testing to protect LLMs from harms like jail-breaking. 5. Tie-ups with top universities for solving hard next generation ML/AI problems. Leading four major innovations, including Scattering Vision Transformer (NeurIPS 2023), SpectFormer (WACV 2025), SiMBA (ICASSP 2025) and Mongoose (ICASSP 2026), are being productised in the commerce AI space of Microsoft. Led element, Walmart’s ML platform and data sciences foundation teams as well as academic collaboration at Walmart Global Tech - solved interesting problems in retail and supply-chain. Building R&D teams from scratch, leading the teams and delivering significant value to stakeholders in solving business problems with AI/ML. Have put many models into production and built platforms used by hundreds of data scientists for building models and delivering enormous value for clients. Saved millions of dollars in cost savings and efficiency improvements with AI/ML. Seven granted US patents and five more in the pipeline. Publications in leading conferences/journals including IEEE transactions and NeurIPS. Speaker at top conferences including NeurIPS 2023, AAAI Industry track 2024, AAAI workshop 2022, CIKM 2021 (https://www.cikm2021.org/accepted-papers/applied), ECML 2020, Strata Data California 2018, Strata Data London 2017, Fifth Elephant 2017 and Fifth Elephant Keynote speech in 2014.

Location

Bengaluru

Work experience

Machine Learning Research Leader at Microsoft, Dec 2021 - Present · 4 yrs 5 mos, Bangalore Urban, Karnataka, India, Head of the Cloud-AI R&D plus AI/ML for commerce platform team in C+E. Vijay’s key contributions to AI R&D include spectral transformers like SpectFormer (WACV 2023), Scattering Vision Transformer (NeurIPS 2023), SiMBA (ICASSP 2025) which have demonstrated that transformers can be implemented efficiently (at much lesser number of parameters) without losing accuracy in computer vision applications such as image classification, image segmentation, object detection as well as time series forecasting. Vijay and his team have productionized several of the above models in commerce for forecasting, anomaly detection and recommendation problems. Vijay has built collaborations both within and outside Microsoft, including ties with academic institutions such as the University of Bath, IIT Bombay, and IIT Madras as well as partnerships with MSR teams in Netherlands and UK.

Director of Data Sciences at Walmart Global Tech, May 2019 - Dec 2021 · 2 yrs 8 mos, Bengaluru Area, India, Heading machine learning platform (element), data sciences foundations and academic collaboration • Headed Data Sciences Foundations team, delivering ML models for multi-modal attribute extraction, incident risk prediction, and CV-based profanity detection. • Led Element.ai platform (with more than 700 users) delivery: o Delivered multi-cloud Delivered multi-cloud ML Model deployment, inferencing and model monitoring for performance, fairness and bias - CPU, GPU, TPU across GCP, Azure and On-prem o Knowledge base and feature store (built on top of Feast) – specific platform features added. o Notebook and IDE based model dev/training  Jupyter, R Studio, Theia, H2O o Delivered AutoML capabilities by integrating with appropriate tools o Launched MLOps features including model monitoring, mean time to deploy models was reduced to 1 hour. o Major use case like ULTRON hyper-parameter tuning platform built on element – improved data scientists productivity by 80% o Another important use case for element – search re-rank pipeline, which improved data scientists productivity by 30%. o Other use case include markdown optimization in International and supply chain last mile transportation. o • Established academic collaborations with Indian universities, kickstarting exploratory research projects with IITM, IITB and IISc. • Spearheaded Data Sciences council initiatives, sharing best practices across hiring, delivery, and publishing.

Senior Director Of Technology at Publicis Sapient, Jun 2017 - Apr 2019 · 1 yr 11 mos, Bengaluru Area, India, Head of data technologies including data engineering, machine learning, deep learning/AI - SapientNitro taken over by Publicis group, combined entity known as Publicis.Sapient.

Director of Technology at SapientNitro, May 2016 - Jun 2017 · 1 yr 2 mos, Bengaluru Area, India, Joining SapientNitro's data team as Director and head of distributed computing practice.

Director and Head, Data Sciences at Clustr, Apr 2015 - May 2016 · 1 yr 2 mos, Bangalore, The amount of business data accessible to Tally in both Indian and global context is enourmous. Adding value to the next generation of Tally by deep analytics of this humongous data is my responsibility.

Education

PhD, Distributed Systems, Indian Institute of Technology, Madras

MS (By Research), Distributed Computing, Indian Institute of Technology, Madras