Sep 2021 – Jun 2024 · 2 yrs 9 mos

• Head of consumer new accounts credit risk models (ADSS) for international markets. Member of Global Decision Science organisation.
• Leading a global functional team (1st line) of 23 colleagues (PhD/Masters), spread across India and Singapore, responsible for model research, implementation, and management of credit risk ML models.
• Collaboration with market risk, central risk strategy, model governance (2nd line), capabilities, and internal audit (3rd line) teams.
• Implemented 13 next generation models and several model adjustments across market portfolios, amounting to net benefit of
• ~$60M over pre-existing model solutions.
• Diligently tracked model performance, studied root cause of model inaccuracies, took timely actions to control credit loss.
• Researched a new technique for building credit risk models, using survival method and Bayesian approach, enabling use of most recent performance data in model development. Most credit risk modeling teams at AmEx, are adopting this new framework for their next generation models.
• Highlighted gaps in the current reject inference techniques used in new accounts. Best practices around reject inference techniques haven’t changed for at least a decade at AmEx. Explored academic research on topics like Self Learning and Siamese Neural Network, expanded use of reject inference methodology to smart stacking framework and closed the gaps in model under-prediction for the marginally approved segment.
• Multiple feature engineering activities including but not limited to – bureau odds index, premium inquiry and trade indices, utilized better bureau tradeline elements by creating time-series based attributes, expanded use of secondary bureau, incorporated home value, address, and banking intelligence across multiple markets.

Apr 2019 – Aug 2021 · 2 yrs 4 mos

• Managed a team of 12 NLP experts. Lead hiring and engagement drives. One of the data stewards for the Global Servicing Group handling responsibilities of managing unstructured data for the organization.
• Implemented complaint identification and categorization model, leveraging call and chat transcripts. The NLP model daily processes a volume of over ~350K US Customer interactions to provide complaint tags and associated categories.
• Automated dispute resolution, by evaluating content in merchant sent copy of receipts and letters, using a customer OCR-NLP BOT hosted through a Netflix Conductor orchestrator. The solution achieved 95% precision on 40% relevant samples, relieving AmEx of $10M in financial exposure, during COVID-19 crisis and a projected impact of $250M at point of arrival.
• Developed 1st generation of VIBES (Valuing Interactions by Evaluating Sentiment) model. Brought different teams together to align on use case - determining incentives for Customer Care Professionals (CCPs). VIBES won the 2023 Edward P. Gilligan Award for Innovation at AmEx.
• One of the leads for enterprise knowledge graphs program, aiding metadata curation, table discovery and join recommendation.
• Partnered with governance, compliance, enterprise digital, model governance (2nd line) and engineering teams.
• Awards: Chairman’s Award (top 1%) at American Express for AY’18.
• Global Servicing Group Tribute Award for “going above and beyond and creating culture of excellence.” Certificate of Excellence for Exceptional Leadership from EVP of Global Servicing Group.

Sep 2017 – Mar 2019 · 1 yr 6 mos

Apr 2016 – Sep 2017 · 1 yr 5 mos

• Business Consulting Group | Advanced Data Science Track
• Beat 5 in-house benchmarks for marketing mix problems and finding optimal sequence of promotions using interpretable deep learning techniques and panel data models
• Detected local healthcare markets based on patient flows in US which highlighted payer-provider dynamics
• Identified leading indicators for study and site enrolment duration for clinical trials
• Inferred if there is a statistical difference in patient persistency for a drug sold across two channels and suggested sales optimisation

Jun 2013 – Dec 2015 · 2 yrs 6 mos · Abu Dhabi, UAE,Oslo, Norway

• Worked with Professor Nils Rudi to set up the Sports Analytics Centre at INSEAD Business School, Singapore, curated course materials and structured assignment modules, and wrote the initial drafts of a soccer analytics book.
• Partnered with soccer analysis firm - Amisco-Prozone, to study real-time positional data and bring up insights for coaching.
• Used player tracking and events data, to generate spatial probability map of teams’ ball possessions.
• Ran massively parallel simulations on GPUs, to generate real-time video, helping coaches review with players, during match time.

May 2012 – Jun 2013 · 1 yr 1 mo · Chennai Area, India

• Built predictive models for product categorization using e-commerce web data
• Automatic extraction of product attributes from web pages for matching of products across different websites

Jan 2008 – Jan 2009 · 1 yr