Meenal Bhatt

Lead Data Scientist·Technical Leader

Mumbai, Maharashtra · Building in public since Jun 2013 (~13.1 yrs)

12+Years in Data Science / ML

10M+Transactions scored monthly

₹40Cr+Fraud losses prevented annually

35%Fewer false positives

Growth trajectory

From analyst to AI decisioning strategy.

A clear shift from running fraud analytics to owning AI-driven risk and decisioning strategy for the business.

Jun 2013Started as a data analyst in BFSI

Mar 2015Shipped first ML fraud model

Dec 2017Owned risk scoring at scale

Mar 2021Moved into InsurTech leadership

Jun 2023Led agentic AI initiatives

Jan 2025Owns AI decisioning strategy

Jun 2013
Started as a data analyst in BFSI
Mar 2015
Shipped first ML fraud model
Dec 2017
Owned risk scoring at scale
Mar 2021
Moved into InsurTech leadership
Jun 2023
Led agentic AI initiatives
Jan 2025
Owns AI decisioning strategy

Signals that matter

Live in Production.

Each signal ties back to a specific, evidenced moment - not a self-rating.

Led3

Model strategy, team direction and cross-functional decisions owned end-to-end.

EvidenceLed a full overhaul of the fraud detection model

Built2

Production ML systems shipped from first line of code to deployment.

EvidenceBuilt a real-time risk-scoring engine for underwriting

Scaled1

Taken from a working model to real transaction volume.

EvidenceScaled claims anomaly detection to 10M+ transactions/month

Solved1

Model failures and drift diagnosed and fixed under real business pressure.

EvidenceSolved a post-COVID model-drift issue in the fraud model

Stack

Daily tools.

What Meenal actually reaches for, pulled from the tags attached to captured moments - not a keyword list.

Work

Key Projects By Role

Case studies, not bullet points - impact numbers attached to each one.

InsureNext AI

Lead Data Scientist · Apr 2021 - Present

5 months

Led a full overhaul of the fraud detection model

Replaced a single monolithic classifier with an ensemble architecture and a feedback loop from investigator outcomes, cutting false positives without slowing down claims.

35% fewer false positives₹40Cr+ fraud prevented annually

Fraud DetectionXGBoostModel Ensembling

Verified via MediumView case study

4 months

Built a real-time risk-scoring engine for underwriting

Built a low-latency scoring API consumed directly at point-of-quote, replacing a rules-only underwriting checklist with a model-backed decision.

Sub-200ms scoring latencydeployed across 3 product lines

Risk ScoringPythonMLOps

Verified via Public ProfileView case study

6 months

Led the build of an agentic underwriting copilot

Combined LLMs, a rules engine and the existing risk models into a copilot that drafts underwriting recommendations for a human to review and approve.

40% faster underwriting turnaround

LLM AgentsRAGLangChain

Self-reportedView case study

3 months

Scaled claims anomaly detection to full volume

Took the anomaly-detection model from a weekly batch job to scoring every claim transaction as it's created.

10M+ transactions/month

Anomaly DetectionScikit-learn

Self-reportedView case study

2 months

Solved a post-COVID model-drift issue

Diagnosed declining fraud-model recall traced to a pandemic-era shift in spending patterns the training data no longer reflected.

Recall restored to 92%

Model MonitoringDrift Detection

Self-reportedView case study

4 months

Built and operated a model governance framework

Set up documentation, explainability reports and approval workflows for every production model to satisfy regulatory review.

Passed 2 regulatory audits

Model GovernanceExplainability

Verified via GitHubView case study

Suvarna Financial Services

Data Scientist · Jun 2013 - Mar 2021Verified · employer record

5 months

Built the first in-house fraud-scoring model

Replaced manual rule-based flagging with a statistical scoring model for retail lending fraud, the NBFC's first ML model in production.

First ML model in production at the org

Fraud ScoringLogistic Regression

Verified · employer recordView case study

4 months

Led the rollout of automated transaction monitoring

Led the rollout of the fraud-scoring model across every retail lending branch, coordinating with ops teams on alert thresholds and escalation paths.

Rolled out across all retail branches

RolloutStakeholder Management

Verified · employer recordView case study

Outside work

What gets built off the clock.

Experiments, writing and talks - separate from paid work, tagged as such.

Open source

shap-report-gen

A small toolkit that turns SHAP explanations into stakeholder-readable PDF reports - built because regulators don't read Python notebooks.

Verified · GitHub★ 210

Article

Why explainability matters more than accuracy in BFSI models

On why a slightly-less-accurate model a regulator can understand often beats a black box that scores higher offline.

Published Aug 20247 min read

Speaker

Building Fraud Models Regulators Can Actually Read

Talk on model governance and explainability practices for fraud and risk models in regulated environments.

Fraud Analytics Summit, Mumbai · Nov 2024

Experiment

LLM-based underwriting explainer

A prototype that turns a model's risk score and top features into a plain-English explanation for underwriters.

Prototype · unshipped

Publication / book

Nothing published yet - this will fill in the moment it happens.

Learning

Formal record.

M.Sc., Statistics
University of Mumbai
2011 - 2013
Certified Analytics Professional (CAP)
Verified · issuer
2019
Deep Learning Specialization
DeepLearning.AI
2022

Captured moments

Not a work log. A record of momentum.

Each shaded day is when Meenal captured something built, solved, scaled or led - density here reflects real activity, not tenure.

LessMore

Open to

Open to the right opportunity.

Roles, speaking, mentoring, or a good conversation - Meenal is listening.

Open to rolesOpen to speakingOpen to mentoringOpen to consultingOpen to lecturing

Connect with Meenal