Headline

Manager, Data Science | Financial Services | Python, ML, GenAI & Causal Inference

About

Manager, Data Science with 10 years of experience in financial services and consulting, focused on fraud & trust domains. As a hands-on player-coach, I architect solutions, write production code, review solution design, and set department-wide data science standards while driving stakeholder adoption and trust. I provide technical leadership, mentoring, and coaching to a pod of 4+ data scientists, building technical depth and driving measurable outcomes. I own the full lifecycle from exploratory analysis and experimental design (hypothesis framing, A/B testing, shadow testing) through production deployment, across imbalanced datasets, behavioural networks, and propensity models. Most of the data science solutions are primarily built in Python using Scikit-Learn and tree-based ensembles (Random Forest, XGBoost), extended with graph analytics for network behaviour and GenAI/NLP (RAG, LLM workflows) for unstructured signals.

Location

Bengaluru

AI stack

anthropic apis

hugging face

langchain

langgraph

Work experience

Staff Data scientist at Standard Chartered, Aug 2024 - Present · 2 y 2 mo, Bengaluru, Karnataka, India, - Hands-on data science lead driving cross-portfolio (xPA) initiatives, mentoring a pod of 4, and shaping department-wide data science standards. —— - Architecting a RAG framework with advanced semantic chunking, optimised vector retrieval, and dual-layer deterministic/semantic guardrails, cutting manual multi-document review time. - Shadow-testing an XGBoost early-alert risk engine as 3rd LoD against first-line business rules, surfacing ~10% more high-exposure exceptions. - Replacing quarterly manual theme reviews with an automated batch risk-scoring engine (LLM-extracted features + tree-based models), compressing detection cycles from quarterly to daily with ~10% higher precision. ——

Senior Data Scientist at Standard Chartered, Dec 2020 - Jul 2024 · 3 yrs 8 mos, Bengaluru, Karnataka, India, - Uncovered 12% more hidden fraud patterns while maintaining stable false-positive rates by building NetworkX topology features and training an XGBoost model to generate high-risk alerts. - Designed and developed client propensity models (Random Forest/LightGBM), confirming a 10% simulated adoption lift for digital nudges via prioritized-targeting backtesting. - Scaled Python based analytics workflows to PySpark across global portfolio of trade and cash clients, cutting data preparation overhead by 60% to drastically accelerate batch execution cycles for proactive risk, fraud and propensity scoring engines.

Data Scientist at HSBC, Jul 2019 - Nov 2020 · 1 yr 5 mos, Greater Bengaluru Area, - To help RMs prioritize outreach to high-value wholesale clients across Middle East & South East Asia, built K-Means clustering combined with ensemble propensity models (Random Forest, LightGBM) to segment and score client potential. - With fragmented client profiles blocking cross-sell visibility across Middle East & South East Asia markets, developed an NLP entity resolution framework (TF-IDF, string-distance, regex, cosine similarity), achieving 95% identity-matching accuracy and unlocking 15% more cross-sell opportunities.

Analyst - Business Intelligence at Accenture, Sep 2014 - Jul 2017 · 2 yrs 11 mos, Gurugram, Haryana, India, - Conducted exploratory data analysis (EDA) to design semantic layers, writing optimised SQL scripts that eliminated data delivery bottlenecks to enable self-service business unit analytics. - Designed and automated interactive BI dashboards for core financial and operational metrics, streamlining executive reporting workflows and reducing manual data compilation overhead.

Education

Master of Business Administration, Business Administration and Management, General, INDIAN INSTITUTE OF TECHNOLOGY KANPUR

B.Tech, Electronics and Communications Engineering, Jaypee University of Engineering and Technology