Mehul Fadnavis

AI & Product Management

Internal AI Data Analyst Agent | Product Owner & Developer

Incubated a self-initiated prototype into a dedicated product, building claude code for analysts in grab. Leading a team of 8, I drive the product strategy, architecture, roadmap planning while being the primary developer.

• Design: Built a Python CLI agent using LangGraph for state persistence focusing on personalised user contexts
• Tooling: Skills-based approach with MCP connections to internal data tools and custom analytics functions
• Evals: Implemented LLM-as-a-Judge & manual eval pipelines to iteratively improve agent accuracy.
• Security: Integrated OAuth to enforce RBAC, ensuring secure access to internal data ecosystems.

Analytics Self-Serve Portal (Founding Team)

• Scale & Impact: Co-developed an Agentic AI-enabled portal now serving 250+ WAUs across the organization
• Infra: Architected scalable, provider-agnostic backend using Google ADK, Docker, and Kubernetes

Instrumentation Automation using LLMs

• Recognition: Won Grab CEO Award (<1% of org) leading 20+ analysts to deploy an automated clickstream tool
• Product: Scoped the MVP, defined user journeys, then built the core LLM pipeline to process unstructured data

Card Offers (Consumer Product)

PM and lead analyst for 0 to 1 product through inception, feasibility, development, awaiting launch

• Role: Acting PM & Lead Dev for 0-to-1 launch, validating feasibility via a self-built React MVP.
• Tech: Engineered an LLM-powered pipeline to scrape & normalize unstructured offer data into a unified schema.
• Product: Owned 0-to-1 roadmap, scoping critical features to navigate strict resource constraints to launch

Machine Learning and Generative AI Initiatives across Grab

• Received Approve 2 rating for a patent filed for SQL optimisation using Large Language Models (LLMs) in Grab
• Chat Data Mining scripts across multiple languages, performed Topic Modeling with LLM interpretation
• Developed LLM based documentation engine to improve documentation coverage and user discoverability
• Developed end-to-end ML models and pipelines for user level lifetime value prediction across all Grab users

Product Analytics

Payments Analytics at Grab (On Platform and Off Platform)

• Strategy: Shaped 40% of H2 2025 roadmap via data discovery, aligning lending & merchant teams on key priorities
• Commercial: Secured buy-in to revive "Shopping Tile," projecting $150K incremental monthly TPV
• Growth: Implemented A/B tests for multiple payment features across diverse touchpoints in the app
• Innovation: Won Finathon ( hackathon) by prototyping a payment feature now scaling to 300k monthly users
• Scale: PIC for 4 key payment flows; engineered Star-schema dashboards handling 100M+ rows to unlock insights
• Created and maintained 15+ data pipelines implementing medallion architecture with 10 silver level tables and 5 gold tables used for product metrics reporting, root cause analysis and dashboarding using PowerBI and Superset

Analytics Consulting

Led commercial strategy for clients in education sector; built internal search tool (Elasticsearch) to unify firm knowledge

Strategic planning for an elite university in Saudi Arabia

• Interviewed 15+ experts across various universities to understand market landscape for various programs
• Performed unmet demand analysis by creating a demand forecast model and market knowledge about supply
• Benchmarked data across top universities in the country to identify success factors and upcoming trends

Transaction Analytics and Consulting

• Built reusable components using Alteryx and Power BI deployed across various use-cases saving delivery time
• Developed search engine using Elasticsearch utilizing old projects allowing for a firm-wide knowledge base

Client facing role, delivering end-to-end ML and Statistical solutions for consumer product verticals

Airlines Predictive Maintenance

• Developed rare event predictive models for a US Airline to identify delays, cut losses and improve fleet efficiency
• Reduced licensing costs and model turnover time by 50% as a result of migration from SAS to R and Pyspark
• Architected user-friendly modeling interface on Azure, enabling non-technical stakeholders to leverage predictive insights and reducing model deployment time from 5 days to 2 days

Sales Forecasting for a major US dairy manufacturing firm

• Developed end to end ML and statistical models for a dynamically updated forecasting tool spanning multiple products across business verticals using Python, Pyspark and HDFS
• Employed multiple statistical (Holt-Winter, Variable Auto Regressor) and machine learning models (Random Forest, Gradient Boosting, Long-Short Term Memory Neural Networks) and picked best performing models
• Deployed integrated pipeline with auto-refresh for forecasts and published to a live Tableau dashboard

Insights Platform: Internal Big Data Capability Development

• Liaised with multiple teams as advisor on big data implementation of advanced analytical projects
• Developed Flask app with a REST API to interface with ServiceNow Ticketing system to streamline support requests
• Implemented Speech to Text in a distributed paradigm using Spark and Azure STT reducing processing time by 80%
• Created ETL scripts to extract data from SQL servers, Graphql APIs and Cloud Data Lakes in 5 engagements

Internal Training Initiatives

• Conducted 2 trainings for new hires (50 attendees) on topics of Data Science like Python, R and Machine Learning

Education and Certification

• AWS Certified Machine Learning – Specialty (MLS-C01) / AWS (2023)
• Nanodegree in ML Devops Engineer / Udacity (2022)
• Masters in Chemical Engineering / IIT Bombay, Mumbai GPA - 8.78/10 (2018)
• Bachelors in Chemical Engineering / IIT Bombay, Mumbai GPA - 8.78/10 (2018)
• Exchange Student, Chemical Engineering / TU Denmark, Denmark GPA - 9.67/12 (2018)