Romil Varadkar
Machine Learning Engineer, Apple
Bengaluru
Summary
- Experienced professional with 13+ years of expertise in developing and implementing machine learning algorithms for extracting valuable business insights.
- Proven track record of integrating Agentic AI solutions into production pipelines for optimal scalability and reliability.
- Published three patents in Graph Neural Network, explainable AI, and Convolutions, demonstrating a dedication to innovation and mastery in the field.
- Proven expertise in Neural Networks, Graph neural networks, Transformers, Sequential models, Tree based algorithms, Forecasting, Regression and unsupervised algorithms.
- Expertise in BigData technologies such as Hadoop and Spark and NoSQL databases such as MongoDB and Cassandra.
Overview
12
12
years of professional experience
5
5
years of post-secondary education
Work History
Machine Learning Engineer
Apple
07.2024 - Current
- Designed, developed & implemented Agentic framework for forecast generation and explainaibility
- Developed forecast tool that autoscales for 20K intersections and generates forecasts under 5 mins
- Designed strategy to evaluate bottlenecks in the supply chain by improving TDM offsets
Lead Machine Learning Scientist
PayPal
04.2021 - Current
- I lead the Payments vertical of the Global Data Science team. I have designed Machine Learning road-map for two successive years within Payments by scoping ML use cases, designing process changes, defining success metrics and setting revenue targets.
- Designed, developed and productionalized 5 Machine Learning solutions to achieve an annualized impact ( reduction) of ~ 25M USD on Transaction Expense.
- Fine-tuned Open AI LLM ( GPT 3.5) to proactively identify and scope business opportunities from Network Bulletins.
- Pioneered Smart transaction Routing to optimize for transaction expense - Developed predictive models to predict cost and probability of success associated with all possible processors and networks to optimize for transaction expense and approval rates.
- Engineered an optimized path for transactions based on high likelihood for refunds, derived from refund prediction ML model.
- Optimized transaction expense associated with Pre-Auth indicators through realtime ML model driven high precision prediction of overcaptures and undercaptures.
- Designed, developed and implemented optimization of acquirer routing for pinless networks.
- Published 3 patents by extending convolutions ( engineered Slope features), reverse engineering Message passing layer of Graph Neural Networks.
Assistant Vice President, Analytics
HDFC Bank
11.2019 - 03.2021
- I lead the Marketing Analytics division for Personal Loan and General Insurance.
- Designed Machine Learning road-map, with objective success metrics for Marketing & Risk analytics within Personal Loan, General Insurance.
- Redesigned the experimentation layer by implementing Multi Armed Bandit to imbibe objectivity and incrementality.
- Developed and productionalized 6 Machine learning models with a combined impact of INR 9Cr.
Staff Data Scientist, Risk
Edelweiss Group
08.2018 - 11.2019
- Designed a road-map to impact all phases of customer life cycle through advanced analytics.
- Machine learning-based acquisition scorecards for Business Loans & Personal Loans - Developed the first customer acquisition scorecard for Business Loans which enabled 'Straight-through' processing of 25% applications without any manual intervention. It lead to reduction in short term delinquency (DPD:30+ in the first 6 months) by 3%.
- Engineered AI-enabled recovery and collection scorecards which helped us pro-actively identify customers most likely to default. Business Impact - Increased recovery by targeted and timely interventions by 9% in a space of 3 months.
Senior Data Scientist, Credit Risk
L&T Financial Services
04.2017 - 08.2018
- Customer Loan approval Scorecard: 2W Loans - Developed Customer acquisition scorecard for Two Wheeler Loans which enabled Straight through processing of 90 % applications without any manual intervention of Credit Underwriters and brought in a reduction in short term delinquency (DPD:30+ in the first 6 months) by 1%.
- Risk-Based Pricing - Improving approval rate came out as a significant driver to increase revenues which in-turn created a need for risk-based pricing models. Interest rate was formulated as a direct function of Expected Credit Risk, which enabled approving high risk customers with higher Rate of Interests.
- ECL - Estimated Credit Loss for NPA Provisioning: Developed models for estimation of PD( Probability of Default), EAD( Exposure at Default) and LGD(Loss Given Default) respectively for Housing, 2 Wheeler and Farm Equipment lending Business.
Data Scientist
Piramal Enterprises
04.2016 - 03.2017
- Image Processing: Implemented a supervised Machine Learning algorithm in R based on optical character recognition to extract digits from a handwritten image.
- Curbing Retailer Churn: Analysis focused on identifying traits within retailer behavior that leads to inactivity in the next 3 months.
- Up-sell: Identifying retailers most likely to Up-sell basis a predictive logistic regression model.
Consultant
Ernst & Young
05.2015 - 04.2016
- Customer Persistency - Life Insurance: Designed, developed and productionalized a ML framework which served as early warning system to identify policies at risk, identify leading causes for attrition and implement levers to increase persistency.
- Attrition Propensity Modelling - General Insurance: Agent Attrition was one of the leading concerns for an Insurance client. Model focused on creating a stringent Attrition definition and using the same to come up with predictive characteristics that best define attrition and hold relevant for a considerable future.
Business Analyst
Yatra.com
01.2015 - 03.2015
- Impact of ECash on driving engagement: Designed an optimum retention strategy to improve the repeat rates and average turn around time per customer.
Associate, Analytics
Axtria
06.2013 - 12.2014
- Predictive modeling to target physicians for an extremely rare neurological disease: The exercise involved identifying patients having a risk of contracting the disease as well as physicians treating such patients.
- Predictive model to compute the loss of market share due to the launch of a competitor drug: Exercise involved identifying physicians at risk of switching over to the competitor drug and the percentage of business at risk for each such physician.
- HEOR Space: Quantify efficacy of a drug to provide clinical and economic evidence through simulation techniques, Markov and DES, implemented from scratch in VBA.
Education
B.Tech - Computer Science & Engineering
Indian Institute of Technology (IIT) Indore
01.2009 - 01.2013
Computer Vision, Nanodegree - undefined
Udacity
01.2019 - 01.2020
Skills
Agents
Natural language processing
Credit underwriting
Supply Chain Optimization
Machine learning
Sequential models
Professional Highlights
- Possess ~11 years of extensive experience in the fields of Data Science, Consulting, and designing Machine Learning road-maps, with a proven track record of success.
- Demonstrates a robust domain expertise in various industries, including Payments, Retail Lending, Credit Risk, Insurance, Fraud, Ecommerce, Pharmaceutical, Marketing, and Product Analytics.
- Proven expertise in Neural Networks, Graph neural networks, Transformers, Sequential models, Tree based algorithms, Forecasting, Regression and unsupervised algorithms.
- Expertise in BigData technologies such as Hadoop and Spark and NoSQL databases such as MongoDB and Cassandra.
- Proven track-record in ML research by publishing 3 patents to create tangible revenue impact.
Timeline
Machine Learning Engineer
Apple
07.2024 - Current
Lead Machine Learning Scientist
PayPal
04.2021 - Current
Assistant Vice President, Analytics
HDFC Bank
11.2019 - 03.2021
Computer Vision, Nanodegree - undefined
Udacity
01.2019 - 01.2020
Staff Data Scientist, Risk
Edelweiss Group
08.2018 - 11.2019
Senior Data Scientist, Credit Risk
L&T Financial Services
04.2017 - 08.2018
Data Scientist
Piramal Enterprises
04.2016 - 03.2017
Consultant
Ernst & Young
05.2015 - 04.2016
Business Analyst
Yatra.com
01.2015 - 03.2015
Associate, Analytics
Axtria
06.2013 - 12.2014
B.Tech - Computer Science & Engineering
Indian Institute of Technology (IIT) Indore
01.2009 - 01.2013
Similar Profiles
Shemeka GloverShemeka Glover
Technical Support Representative at AppleTechnical Support Representative at Apple
Rohn BeardRohn Beard
Inside Business Expert (RCC) at AppleInside Business Expert (RCC) at Apple
Jonah LoucksJonah Loucks
Social Media Specialist at AppleSocial Media Specialist at Apple
Kaiden EarlKaiden Earl
Technical Specialist at AppleTechnical Specialist at Apple
Prabhanjan PadakandlaPrabhanjan Padakandla
Client Services Lead Associate at Yahoo! Inc.Client Services Lead Associate at Yahoo! Inc.