Jan 2025 – Sep 2025 · 8 mos · San Francisco Bay Area · On-site

• As the founder of Peekly, built an AI-driven marketing analytics platform from the ground up. My vision was to enable a truly conversational "talk to your data" experience, empowering businesses to get instant insights without complex dashboards.
• Designed and built a unified, multi-agent LangGraph pipeline to integrate and analyze siloed data sources like web traffic, ads, e-commerce, and payments. This approach slashed the time it takes to get actionable insights from approximately two hours to under 30 seconds.
• Crafted seamless API integrations with platforms like Google Analytics, Shopify, and Stripe, which automated data synchronization and reduced client onboarding time from days to just minutes.

Jan 2022 – Jul 2024 · 2 yrs 6 mos

• Top Impactor 2024
• Spot awards 2024, 2023 and 2022
• Transforming Businesses with AI & Data
• Worked as an accomplished leader and builder, specializing in leveraging advanced data analytics and AI to create strategic solutions that drive business transformation. Led diverse, cross-functional teams to develop innovative systems that boost efficiency and profitability, from demand forecasting to root cause analysis.
• At Converge Consumer, I led a 12-member team to deliver Demand Brain, a custom-featured forecasting engine that boosted accuracy by 15%, directly optimizing marketing, pricing, and promotional activities. I engineered a PySpark pipeline and a Feature Factory using Kubeflow and Feast to automate feature engineering, slashing analysis time by over 90% and development time by 75%.
• As part of the Smart Factory initiative, I developed a chat-based Generative AI multi-agent system for root cause analysis in manufacturing. This system, using a RAG framework, diagnosed faults by analyzing sensor data and manuals via a knowledge graph, accelerating identification by 80% and automating ServiceNow ticketing. Through A/B testing, DeepEval, and Pairwise Preference Testing, I improved the LLM's response approval rate by 42%.
• During my work with Google, I built Cymbal Labs, a cloud-based enterprise search solution using LangChain, Vertex AI, and Search API, showcased at Google Cloud Next ’23. I optimized the system’s performance by reducing latency from 6.0s to 2.2s and introduced a converse mode with Redis-backed session memory for persistent, multi-turn interactions.
• My passion is turning complex data and AI challenges into tangible business advantages, focusing on technical excellence that delivers measurable impact.

Jun 2021 – Jan 2022 · 7 mos

• Specialized AI & NLP Solutions
• As an accomplished developer and builder with a passion for using Natural Language Processing (NLP) and AI to solve complex, domain-specific challenges. Worked on solutions that transform how information is accessed, analyzed, and understood in specialized fields like legal and medical research.
• Key Projects
• PubMed Search: I developed a specialized NoSQL data store solution on DynamoDB to enable context-based, zero-shot search across vast amounts of medical research data. This system leveraged AWS Comprehend's Named Entity Recognition (NER) to intelligently parse and organize gigabytes of information, making it more accessible and useful for researchers.
• Legalisis: I architected a contract comparison system that cut down legal review time from hours to minutes. This solution used Sentence-BERT for semantic similarity and SpaCy NER for key entity extraction. I also developed an abstractive summarization model and a one-shot BERT model to create a comprehensive legal document analysis platform. This platform provides rapid insights and identifies critical terms and conditions, streamlining the legal review process.

Feb 2021 – Jun 2021 · 4 mos

Jun 2020 – Jan 2021 · 7 mos · Pune, Maharashtra, India

Jun 2020 – Aug 2020 · 2 mos

• Worked on a project to autogenerate C++ code from truth tables, in SoP format.

May 2020 – Jul 2020 · 2 mos

• Developed a supervised model relating the spread of covid-19 with the data released annually by UNDP for the human development index and data from WHO.

Feb 2020 – Jun 2020 · 4 mos

• Enhancing Market Intelligence with Data Science
• Worked as a Data Science Consultant at Liquid Diamonds, where I leveraged advanced analytics to build solutions that enhanced market intelligence and improved platform liquidity.
• Forecasting & Recommendations: I developed a time series forecasting API using ARIMA to model diamond market trends. This solution generated personalized buy/sell recommendations with an 82% accuracy, providing valuable insights into market liquidity for users.
• Pricing & Anomaly Detection: I also built a tree-based regression pricing API that achieved an R-squared score of 0.95. This API enabled standardized diamond price predictions and was used to automatically flag pricing anomalies across the platform, ensuring greater market transparency.

Jul 2019 – Aug 2019 · 1 mo · Israel

• Participated in Data Mining & Business Intelligence for Cyber Security program
• 2019.
• Presented my research on detecting Trojan attacks on Neural Networks Models
• using the SHaP algorithm.

May 2019 – Jul 2019 · 2 mos · New Delhi Area, India

• Automating Cybersecurity with Machine Learning
• As an intern at QOS Technology, I gained hands-on experience applying machine learning to cybersecurity challenges. I developed an anomaly-detection model designed to automate a significant portion of a cyber analyst's workflow.
• My work focused on analyzing SIEM log data to identify complex, multi-step cyber threats, specifically 7-step kill-chains. The model was built using advanced ensemble ML techniques, including genetic algorithms for optimization and LSTM networks for analyzing sequential log data. This project provided valuable insights into using AI to enhance cybersecurity defenses and streamline threat detection processes.

Dec 2018 – May 2019 · 5 mos · Gurgaon, India

• Audience Insights & Content Strategy
• As a Data Analyst at Times Internet, I focused on leveraging data to uncover audience insights and drive content strategy. My work contributed to increasing content engagement and virality across a large portfolio of digital properties.
• Audience Segmentation: I built segmentations using K-means clustering on user behavior and overlap across more than 40 Times Group companies. This analysis identified 12 distinct audience archetypes, which were shared across the entire portfolio.
• Flywheel Framework: I used these audience similarity insights to devise a flywheel framework. By understanding how audiences moved between different platforms, It helped increase content virality metrics by 18%.

Dec 2018 – Jan 2019 · 1 mo · Bennett University

• Mentored 4 interns to develop a CNN-based Road Assets Detection System.
• The network can detect potholes, vehicles and traffic signs with an accuracy > 84%

Sep 2018 – Nov 2018 · 2 mos

Jul 2018 – Mar 2020 · 1 yr 8 mos

Sep 2018 – Jul 2019 · 10 mos

Aug 2018 – Sep 2018 · 1 mo

Jun 2018 – Aug 2018 · 2 mos

Jun 2018 – Jul 2018 · 1 mo · Greater Noida

Aug 2017 – Aug 2018 · 1 yr

Aug 2017 – Aug 2018 · 1 yr