Career & Projects
A decade building AI that ships
12 years across search, recommendations, and agentic systems — leading teams, owning platforms, and turning models into products that move real metrics.
Experience
Where I've made an impact
Building enterprise AI products and leading high-performing teams across three companies.
Gurgaon · Jun 2018 – Present
GARTNER
Data Scientist → Sr. Data Scientist → Director → Sr. Director (4 promotions)
Led the Gartner Digital Experience India data science team with Product and IT to build end-user AI products, overseeing 7 scrum teams of 40+ data scientists, data engineers and engineers. Built AskGartner, Gartner's GenAI advisor, with LangGraph agents, RAG over Gartner content and a metrics-driven evaluation framework, raising answerability to 95%. Managed Gartner.com Search, a $300M product: hybrid search, the SOLR → OpenSearch migration and personalized rankers drove a 10%+ QAR lift. Built Gartner's first conference recommender (90%+ attendance rate) and rolled out MLOps/LLMOps. Earlier, as a data scientist, built client-risk and retention products and grew the team from 3 to 17.
Gurgaon · Mar 2017 – Jun 2018
AMERICAN EXPRESS
Developed scalable ML models for online credit card applications to increase customer acquisition. Led modeling and analytics for international credit card acquisition through digital platform.
Mumbai · Jun 2014 – Mar 2017
GEP Solutions
Developed patented Spend Data Classification tool using Big Data & AI, reducing delivery times by 80% with 30% accuracy uplift. Built proprietary search engine on Elasticsearch for 300Mn+ line items.
Key projects
AI products that drove impact
Measurable business outcomes across agentic AI, search, recommendations, and classification.
AskGartner — Agentic Conversation Product
Gartner
Gartner's own GenAI advisor — an agentic conversational product that answers client questions grounded in Gartner research, built on LangGraph agents with frontier LLMs (GPT, Claude, Gemini) and in-house trained models.
Key achievements
- Raised answer quality and answerability to 95% with loopy agents and LangGraph deep-agents
- Connector tools, RAG and search pipelines that ground the agent in all Gartner assets
- Evaluation framework that ties development, testing and release to metrics
- Adoption-led rollout focused on coverage, stickiness and quality
Technologies
Enterprise Search Platform
Gartner
Owned Gartner.com Search, a $300M product — re-architected the query pipeline to be ML-ready and experimentation-friendly, reusable across every searchable Gartner asset.
Key achievements
- Led the SOLR → OpenSearch migration with hybrid search and SOTA retrieval + ranking (10%+ QAR uplift)
- 50% lower search latency via a Python query-pipeline rewrite and model optimization
- Personalized, low-latency typeahead — 50% uplift in usage, 10%+ content CTR
- Deep-learning document and multimedia search with personalized re-ranking
Technologies
Conference Recommendation System
Gartner
Deep learning recommendation system for 30+ annual conferences.
Key achievements
- Real-time recommendations via API
- Automated retraining with drift detection
- Integrated into Salesforce for sales teams
Technologies
Spend Data Classification (Patent)
GEP Solutions
AI-powered automated tool to classify enterprise spend data.
Key achievements
- US Patent granted (US-10438133-B2)
- Elasticsearch-based search on 300M+ line items
- Production system processing billions in spend data
Technologies
Open source
Open Source
Tools I build and share with the community.
secscan
AI/ML-aware security scanner
A self-contained scanner that orchestrates OSS tools across code, APIs, cloud and AI/ML pipelines — normalized to OWASP / MITRE ATLAS / CWE.
ds-llm-eval
Unified evaluation toolkit for search, recsys & LLM/agents
One pip-installable framework to score retrieval, recommendation, and LLM/agentic pipelines through a single API — ranking, beyond-accuracy, text, LLM-judge and agentic metrics, with significance testing, YAML-reproducible runs and Langfuse logging.
Tech stack
Tools for production AI systems
The technologies I reach for when building and shipping AI products.
GenAI, Agents & Search Engine
Machine Learning and Data Science
MLOps & Cloud
Data Engineering
Programming & Tools
Analytics & BI
Education
Foundations
Statistics and applied math from two of India's leading institutions.
2012 – 2014
IIT Bombay
Master of Science · Applied Statistics & Informatics
2009 – 2012
University of Delhi
Bachelor of Science (Hons.) · Statistics