About
I build AI systems that work in production, not just in notebooks. Over 9 years, I've worked across banking and media, shipping GenAI, RAG, and recommendation systems that drive measurable outcomes. At Wells Fargo, I led a RAG-based virtual assistant that indexes 600+ banking documents and achieves 95% search accuracy, with a patent pending on the architecture. At Zee Entertainment, I worked on knowledge graph-based search and recommendation systems that grew effective content catalog reach from 65 to 430 titles and contributed to a 9.6% watch-time uplift for 2M+ users. My experience spans LLM fine-tuning, agentic pipelines, vector search, MLOps, feature stores, and model governance. I enjoy working across the full cycle from research and prototyping to production deployment and team enablement. Outside of work, I speak at AI events and sessions, enjoy mentoring teams, and like breaking down complex topics for practical audiences.
Work experience
Director - Data Science at Eloelo, May 2026 - Present · 4 mos, Bengaluru, Karnataka, India
Vice President - Generative AI at Wells Fargo, Feb 2026 - May 2026 · 4 mos, - Leading GenAI strategy and architecture for banking workflows, with a focus on production LLM systems at enterprise scale. - Working across AI, engineering, compliance, and business teams to embed GenAI capabilities into core banking operations. - Expanding the RAG platform to support agentic workflows using Google ADK, beyond document retrieval. - Mentoring AI engineers and data scientists on LLM evaluation, prompt engineering, and responsible deployment practices.
Senior AVP - Generative AI at Wells Fargo, May 2024 - Feb 2026 · 1 yr 10 mos, - Built a RAG-based banking virtual assistant achieving 95% Hit Rate@3, indexing 600+ documents with HTML-aware chunking and custom embedding models, reducing resolution time by over 70%. Patent pending. - Designed a validation pipeline generating 7,000+ QA pairs using LLM-as-Judge with human-in-the-loop review to ensure production reliability. - Maintained LLM Ops infrastructure covering model evaluation, retrieval benchmarking, and improvement cycles. - Led cross-functional integration efforts to connect the AI system with existing banking workflows.
Senior Data Scientist at Zee Entertainment Enterprises Limited, Apr 2021 - Mar 2024 · 3 yrs, Bengaluru, Karnataka, India, - Worked on India's first production music generation model, enabling text-to-music creation bridging Indian and Western instruments, built on Meta's MusicGen with a 5-member team. - Delivered Zee's first in-house recommendation model using LightFM, combining collaborative filtering and content-based features for personalised discovery, growing effective catalog from 65 to 430 titles and contributing to a 9.6% watch-time uplift across 2M+ users. - Built a conversational recommendations bot using ChatGPT 3.5 Turbo for multi-catalog user intent understanding.
Data Scientist at Zee Entertainment Enterprises Limited, Jul 2020 - Apr 2021 · 10 mos, Mumbai, Maharashtra, India, - Deployed an AutoML-based batch prediction model for personalised CLM outreach, contributing to an 80% increase in subscriber conversions. - Built time-series forecasting and decay models for subscription and watch-time trajectories, supporting scenario planning that contributed to a 2% revenue lift. - Created demand elasticity models for pack pricing optimisation, contributing to 12% growth in customer LTV. - Automated ETL and BI workflows, saving 30+ human hours daily.
Data Science Intern at Zee Entertainment Enterprises Limited, Jan 2020 - Jun 2020 · 6 mos, Mumbai Area, India, - Built anomaly detection and forecasting models that reduced average platform downtime from 6 hours to 1 hour, with real-time multi-threshold alerting.
Data Science Intern at Siemens Healthineers, May 2019 - Jul 2019 · 3 mos, Bengaluru Area, India, - Designed a scalable ETL pipeline to process global ultrasound machine logs, extracting structured data from ZIP archives without full decompression. - Automated log parsing and storage in MongoDB, enabling interactive insights into probe usage by country, region, and machine type. - Built an XGBoost classification model for predictive maintenance, achieving 83% accuracy and contributing to a 10% reduction in machine downtime.
Social Media Analyst at XOR Labs, Mar 2017 - Jul 2018 · 1 yr 5 mos, Mohali, Punjab, India, - Built a social media ROI analysis framework to measure campaign performance across IPL and global sporting events. - Developed predictive models for audience engagement, contributing to published industry benchmarks in sports analytics. - Combined data analysis with marketing strategy to deliver actionable insights for brand and campaign decisions.
Cofounder at GharTool, Mar 2016 - Feb 2017 · 11 mo, Bathinda
