---
title: "Dhaval Thakkar - Lead Data Scientist | Product & Delivery Lead in Mumbai | Tal"
description: "Meet Dhaval Thakkar, Lead Data Scientist & AI Delivery Lead | Building Production AI Systems based in Mumbai. AI stack includes claude, codex, and cursor."
image: "https://tal.club/meet/dhaval-thakkar/share-image"
---

# Dhaval Thakkar on Tal

## Headline

Lead Data Scientist & AI Delivery Lead \| Building Production AI Systems

## About

I build production AI systems that connect LLMs, retrieval, data, evaluation, observability, and business workflows into products people actually use\. I am a Lead Data Scientist and AI Delivery Lead with close to 8 years of experience across enterprise agentic AI, RAG systems, production ML, MLOps, data engineering, and product delivery\. Currently, I work through Spinnaker Analytics on enterprise AI products for John Hancock Finance, where I own the full delivery lifecycle across architecture, product strategy, development, testing, DevOps, stakeholder management, and production operations\. My recent work includes building and scaling an agentic AI assistant used by 1,500\+ daily users\. The platform helps sales and business teams answer complex questions across underwriting guidelines, policy configurations, premium\-related workflows, enterprise documents, and analytics systems\. It combines LangGraph\-based orchestration, specialist agents, Self\-RAG, Text\-to\-SQL, vector search, middleware controls, observability, LLM evaluation, and safe fallback behavior for regulated financial services workflows\. Earlier, I worked across quick commerce, enterprise analytics, retail loyalty, logistics, OCR, dynamic pricing, customer segmentation, and experimentation\. At Zepto, I built data and experimentation systems that supported loyalty and retention programs at national scale\. At Emplay Analytics, I helped build an MLOps framework that reduced delivery effort and became the company’s standard blueprint for analytics engagements\. At Unifynd, I built ML and OCR systems deployed across major Indian retail properties including Phoenix Market City and Palladium\. I am most effective in roles where I can own the full delivery chain: understand the business problem, design the system, build the core components, lead the team, manage stakeholders, and make sure the product is adopted in production\. Core areas: Agentic AI, RAG, LangGraph, LangChain, OpenAI, Anthropic, Azure OpenAI, vector databases, Qdrant, LangFuse, FastAPI, Python, SQL, MLOps, MLflow, Kedro, Great Expectations, dbt, Docker, Kubernetes, Terraform, Azure, AWS, data engineering, experimentation, and production ML\.

## Location

Mumbai

## AI stack

claude

codex

cursor

deepseek

hermes agent

langgraph

## Work experience

Lead Data Scientist \| Product & Delivery Lead at Spinnaker Analytics, Feb 2025 \- Present · 1 yr 7 mos, Engaged with John Hancock Finance through Spinnaker Analytics to build and scale enterprise AI products for financial services\. Own the full delivery lifecycle across architecture, product strategy, development, testing, DevOps, stakeholder management, and production operations\. Lead cross\-functional delivery across engineering and business teams\. Key work: • Architected and delivered an enterprise agentic AI assistant using LangGraph, specialist agents, RAG, Text\-to\-SQL, vector search, and safe fallback workflows\. • Built a routed agent architecture that decomposes complex questions and sends them to the right specialist capability, including Self\-RAG, SQL\-based analytics, and general fallback flows\. • Designed the retrieval foundation using document processing, metadata extraction, embeddings, and vector indexing across enterprise documents, sales collateral, underwriting material, policy content, and legacy knowledge sources\. • Integrated observability and evaluation workflows using LangFuse, production tracing, prompt versioning, cost visibility, and LLM\-based evaluation on real user traces\. • Built middleware controls for routing, retries, grounding checks, fallback behavior, and safer response generation in regulated insurance workflows\. • Helped compress query turnaround from days to seconds for sales and business teams\. • Scaled the AI assistant to 1,500\+ daily users, with an estimated $5–7M annualized productivity benefit based on measured workflow time savings\. • Extended the platform into SEC 10\-K financial analysis, enabling analysts to compare insurance carriers across filings, metrics, peer rankings, CAGR, YoY calculations, and multi\-company search in seconds\. Tech: Python, FastAPI, LangGraph, LangChain, Qdrant, Azure OpenAI, Anthropic, OpenAI, LangFuse, Docker, Azure, GitHub Actions, Terraform, Azure Key Vault\.

Senior Data Scientist at Spinnaker Analytics, Dec 2022 \- Jan 2025 · 2 yrs 2 mos, • Worked on enterprise AI, retrieval systems, document intelligence, and production ML solutions for financial services\. • Built LLM\-powered workflows for enterprise knowledge retrieval across underwriting, sales, policy, and operational documents\. • Designed text processing and embedding pipelines to convert PDFs, legacy documents, and unstructured enterprise content into searchable knowledge bases\. • Developed RAG workflows with vector databases, metadata filtering, prompt orchestration, and evaluation loops\. • Worked with OpenAI, Anthropic, Azure OpenAI, LangChain, LangGraph, Qdrant, MLflow, Python, FastAPI, Docker, and Azure services\. • Collaborated with product, engineering, business, and stakeholder teams to move AI use cases from prototype to production\.

Senior Manager \- Retention Marketing at Zepto, Jul 2022 \- Dec 2022 · 6 mos, Mumbai, Maharashtra, India, • Owned experimentation, data engineering, and operational analytics for Zepto’s 10\-minute delivery platform, working with product and operations leadership\. • Built data pipelines and experimentation systems for loyalty and retention programs across India\. • Designed A/B experiments for cashback incentives, user\-specific offers, and loyalty mechanics\. • Supported a controlled loyalty pilot deployed to 10% of users across India, resulting in a 33% increase in order frequency compared to the control group\. • Built an issue identification framework across store, city, and user levels to detect operational bottlenecks affecting retention\. • Consolidated data from Mixpanel, CleverTap, warehouse systems, and internal operational sources\. • Built diagnostic views used by store managers, city teams, and customer support teams to proactively identify issues\. • Contributed to a 14% improvement in user retention through operational diagnostics and retention analytics\.

Product Manager at Emplay Inc\., Oct 2021 \- Jul 2022 · 10 mos, Mumbai, Maharashtra, India, Led analytics product delivery and MLOps transformation for enterprise clients including Google, SAP, Cisco, and other large organizations\. • Led an 8\-person cross\-functional team delivering sales deal analysis and ML solutions for enterprise clients\. • Built and deployed ML models to predict sales deal status with 86% accuracy\. • Designed an end\-to\-end MLOps framework using Kedro, Great Expectations, dbt, and MLflow\. • Standardized data ingestion, preprocessing, data quality checks, model tracking, deployment, monitoring, and scheduled runs\. • Reduced delivery turnaround from 3 months to 1 month\. • Reduced per\-client delivery effort from 15–20 people to 4–5 people by converting manual delivery into a repeatable, automated framework\. • Helped establish the framework as the standard blueprint for future data science projects\.

Data Scientist at Unifynd Technologies Pvt\. Ltd\., Jun 2018 \- Oct 2021 · 3 yrs 5 mos, Mumbai, Maharashtra, India, • Built and scaled ML, OCR, pricing, segmentation, and reporting systems across retail loyalty, malls, and logistics\. • Built a templatized OCR engine with GUI to verify customer receipts across 5,000\+ retailers\. • Designed extraction, confidence scoring, fraud detection, and human review workflows for receipt verification\. • Deployed the OCR platform across major retail properties including Phoenix Market City, Palladium, DLF Mall, and other mall networks\. • Built customer segmentation models that increased revenue by 3x and coupon utilization by 2x\. • Built a dynamic coupon pricing engine that increased coupon consumption by 5x\. • Built pricing and customer segmentation models for a logistics company, reducing customer attrition by 32% and increasing repeat purchase rate by 12%\. • Automated KPI reporting across 10 malls, saving 60\+ hours of manual effort per month\.

Intern Data Scientist at Unifynd Technologies Pvt\. Ltd\., Sep 2017 \- Jun 2018 · 10 mos, Mumbai, Maharashtra, India, • Developed Flask based middleware APIs for interfacing backend data science algorithms with frontend • Built a smart cryptocurrency trading platform which used social data and historical prices to optimize current portfolio\. Boosted overall profit from the portfolio by 30% • Worked with Product and Marketing teams to identify the power users of an app which resulted in 43% increase in activity and a 65% increase in revenue from these users

Deep Learning Intern at ZIFF, Inc\., May 2017 \- Aug 2017 · 4 mos, United States, • Machine Learning internship with an emphasis on Deep Neural Networks • Demonstrated competency in Hyper\-parameter optimization, Image Augmentation, and Learning rate Decay strategies on huge Image datasets • Worked with Keras, Python and Docker

## Education

Bachelor of Engineering, Computer Engineering, University of Mumbai

[View the original profile](https://tal.club/meet/dhaval-thakkar)

```json
{
  "@context": "https://schema.org",
  "@graph": [
    {
      "@type": "Person",
      "@id": "https://tal.club/meet/dhaval-thakkar#person",
      "name": "Dhaval Thakkar",
      "url": "https://tal.club/meet/dhaval-thakkar",
      "image": "https://tal.club/meet/dhaval-thakkar/profile-photo.jpg",
      "description": "Meet Dhaval Thakkar, Lead Data Scientist & AI Delivery Lead | Building Production AI Systems based in Mumbai. AI stack includes claude, codex, and cursor.",
      "jobTitle": "Lead Data Scientist | Product & Delivery Lead",
      "worksFor": {
        "@type": "Organization",
        "name": "Spinnaker Analytics"
      },
      "hasOccupation": {
        "@type": "Occupation",
        "name": "Lead Data Scientist | Product & Delivery Lead"
      },
      "homeLocation": {
        "@type": "Place",
        "name": "Mumbai"
      },
      "alumniOf": [
        {
          "@type": "EducationalOrganization",
          "name": "University of Mumbai"
        }
      ],
      "knowsAbout": [
        "claude",
        "codex",
        "cursor",
        "deepseek",
        "hermes agent",
        "langgraph"
      ],
      "mainEntityOfPage": {
        "@id": "https://tal.club/meet/dhaval-thakkar#webpage"
      }
    },
    {
      "@type": "ProfilePage",
      "@id": "https://tal.club/meet/dhaval-thakkar#webpage",
      "url": "https://tal.club/meet/dhaval-thakkar",
      "name": "Dhaval Thakkar on Tal",
      "description": "Meet Dhaval Thakkar, Lead Data Scientist & AI Delivery Lead | Building Production AI Systems based in Mumbai. AI stack includes claude, codex, and cursor.",
      "dateModified": "2026-08-02T02:11:55.083Z",
      "isPartOf": {
        "@id": "https://tal.club/#website"
      },
      "mainEntity": {
        "@id": "https://tal.club/meet/dhaval-thakkar#person"
      },
      "primaryImageOfPage": {
        "@type": "ImageObject",
        "url": "https://tal.club/meet/dhaval-thakkar/profile-photo.jpg",
        "contentUrl": "https://tal.club/meet/dhaval-thakkar/profile-photo.jpg",
        "caption": "Dhaval Thakkar profile photo on Tal"
      },
      "inLanguage": "en-IN"
    }
  ]
}
```