Headline

SVP Technology at GeoIQ (a Lenskart company) | Production AI Agents | Demand Forecasting, Replenishment & Assortment ML | AI & Location Intelligence for Retail | IIT Bombay

About

I lead technology at GeoIQ, where we build AI and location-intelligence systems that decide where retailers open stores, what they stock, how they replenish, and how their customers shop. Since Lenskart acquired GeoIQ in 2025, much of that work runs inside Asia's largest eyewear retailer. Three things I've spent the last few years on: Store expansion. Our ML models analysed 3,000+ location attributes to guide Lenskart's store launches, cutting non-performing stores by half. The same stack powers GeoIQ's RetailIQ, MarketConnect and PropIQ products for other retail brands. Merchandising intelligence. I architect and run a demand-forecasting, inventory-replenishment and assortment-planning platform for Lenskart's retail network — global tree-based models and probabilistic forecasting, intermittent-demand methods and hierarchical reconciliation that turn a forecast into an actual buy decision. Production AI agents. I've taken LLM-based agents from demo to production at Lenskart scale — most visibly 'B', the visual AI stylist live in the Lenskart app that changes how customers shop for eyewear online. The unglamorous part is what makes it work: evaluation, observability, failure-mode handling and the feedback loops that keep agents reliable once real users arrive. Under the hood: GCP, Kubernetes, BigQuery, Python, and a lot of opinion about data pipelines. Before GeoIQ I co-founded TorcAI, where I built the ML behind a programmatic ad exchange and DSP and a content-recommendation system serving 60M+ users. Earlier I set up data science at JetSynthesys and built recommender systems for Shine.com at HT Media. Dual degree in Aerospace Engineering from IIT Bombay. Happy to talk retail ML, forecasting that survives contact with a store, or making AI agents boring enough to trust.

Location

Bengaluru

Work experience

Senior Vice President of Technology at GeoIQ, Jun 2023 - Present · 3 yrs 4 mos, Bengaluru, Karnataka, India, Lead engineering, data and ML at GeoIQ, a location-intelligence company acquired by Lenskart in 2025 to build its AI capabilities. • Architected Lenskart's merchandising intelligence platform: demand forecasting, inventory replenishment and assortment planning across the store network, built as modular services on GCP/Kubernetes with BigQuery as the analytical backbone. • Built and shipped production LLM agents for Lenskart, including 'B', the visual AI stylist live in the Lenskart app; own the evaluation, observability and feedback-loop tooling that keeps them reliable at scale. • Continue to scale GeoIQ's location-AI products (RetailIQ, MarketConnect, PropIQ) used by retail brands for site selection, revenue estimation and rental benchmarking. • Grew the tech, data and product teams through the acquisition and integration with Lenskart.

VP of Technology at GeoIQ, Jul 2021 - Jun 2023 · 2 yrs, Bengaluru, Karnataka, India, Owned the technology and data-science function for GeoIQ's location-intelligence SaaS. • Led the ML work behind Lenskart's offline expansion: models over 3,000+ location attributes that guided store launches across India and halved the rate of non-performing stores. • Built GeoIQ's core location-data platform — footfall and mobility signals, POI data and revenue-estimation models for prospective sites. • Scaled the engineering team roughly 2x in line with the company's FY23 growth plan.

Co-Founder & Head of Analytics at TorcAI Digital Media Private Limited, Jul 2017 - Jul 2021 · 4 yrs 1 mo, Pune/Pimpri-Chinchwad Area, Co-founded an AdTech company and built its data science function from zero. • Designed and built ION Exchange, an AI-driven programmatic ad exchange that monetised publisher inventory to maximise fill and revenue; the surge-pricing model lifted publisher revenue 10–15%. • Built ION Bidder, a programmatic DSP with deep-neural-network CPM/CPC/CPA bidding models; handed over to multiple clients at a 20–25% revenue uplift over their incumbent systems. • Built a content-recommendation system for one of India's largest news publishers (60M+ users, 10M+ articles). • Architected the low-latency, high-throughput real-time-bidding stack on Spark, Kafka, Aerospike and HBase.

Lead Data Scientist at JetSynthesys Pvt. Ltd., Jan 2015 - Jun 2017 · 2 yrs 6 mos, Pune, Maharashtra, India, Set up and led the data science team. • Built a user-recommendation engine for programmatic targeting on the IntelliExchange ad platform — 30% CTR lift, 2x revenue. • Processed ~1 TB/day of CRM and network data on a MapR Hadoop cluster (Spark, Hive, MLlib, Kafka); built near-real-time URL categorisation and user-segmentation taxonomies. • Client analytics: inbound-email topic classification and sentiment for a leading Indian bank (LDA); bill-shock analysis and CDN caching strategy for a Nigerian 4G ISP.

Data Scientist at HT Media Ltd, Dec 2013 - Jan 2015 · 1 yr 2 mos, Gurugram, Haryana, India, Recommendation systems and lead scoring for Shine.com and HTCampus.com. • Built Similar Jobs and Similar Candidates recommenders (hybrid collaborative + content-based filtering) on Hadoop / Amazon EMR; targeted job-recommendation emails raised applications 25% and logins 20%. • Logistic-regression lead scorecard that prioritised tele-verification and improved lead conversion 20%.

Senior Software Developer at Mphasis, Jul 2011 - Nov 2013 · 2 yrs 5 mos, Mumbai Metropolitan Region, Banking software for Misys Loan IQ and a liquidity-risk analysis tool. • Built a loan-migration automation utility (CIMS → Loan IQ), rolled out for the Malaysia and Singapore businesses. • Liquidity Risk Infrastructure Analysis Tool: daily RDBMS-to-HDFS data migration and client risk reporting on archived data.

Education

Dual Degree (BTech.+ MTech.), Aerospace Engg., Indian Institute of Technology, Bombay