Shivam Arora

Experience

• 

  Indian Institute of Management, Ahmedabad

  May 2013 - Jun 2013

  Summer Internship

  Evaluated ideas/innovations received from grassroots innovators, analyzed trend of entries received from two states in Ignite'13, visited innovator's place for detailed documentation and did prior art search of grassroots technologies .

• 

  Indian Institute of Technology (Banaras Hindu University), Varanasi

  Sept 2013 - Mar 2015

  Student Manager, Entrepreneurship Cell

  Organized workshops, case study competitions, and startup-focused events to foster innovation and experiential learning.

• 

  Fractal Analytics

  Jun 2015 - May 2016

  Business Analyst

  Analyzed & tracked global pricing & promotions strategy of 66 brands, 200+ products for Fortune100 CPG firm

• 

  PwC Analytics Advisory

  Oct 2017 - Apr 2018

  Intern

  - Implemented Sentiment analysis and Topic Modeling on twitter and news feed using Gensim and NLTK library- Developed ‘5G demand forecast’ model to assist telecom clients in resource planning, in collaboration with PwC USA- Segmented customers based on geo-demographic data using K-means Clustering; profiled segments using business logic

• 

  Piramal Group

  May 2018 - Aug 2020

  Analytics Manager

  Piramal Capital & Housing Finance operated as a startup within the broader Piramal Group. I led 0-to-1 data science initiatives, collaborating with cross-functional internal teams and external partners to deliver high-impact solutions. -----------------------------------------------Account Monitoring – ML for Risk Management----------------------------------------------- - Built & deployed a delinquency prediction model using Random Forest + ADASYN, improving GINI by 45%. - Monitored performance via PSI, GINI drift, and accuracy. - Evaluated third-party datasets to enhance model robustness.-----------------------------------Risk Analytics – Home Loan & LAP----------------------------------- - Developed underwriting & reject inference models with Experian Bureau for sourcing - Built executive Power BI dashboards for risk & collections. - Assessed DHFL portfolio risk for acquisition purpose - Leverage K-means on bureau data to segment leads on risk profile and propensity to take home loan-------------------------------------------------------Customer Journey – Marketing & Funnel Optimization------------------------------------------------------- - Defined digital KPIs across marketing, sales & support to improve efficiency. - Ran in-house campaigns across Google, Facebook & Bing, reducing CoA by 45–55%. - Built keyword classifiers & lookalike models to optimize targeting. Show less

• 

  Aera Technology

  Sept 2020 - Oct 2022

  Senior Data Scientist

  ---------------------------------------------------------------------Smart Skill – Reinforcement Learning for Automated Decision Making---------------------------------------------------------------------- Enabled self-learning in the Aera platform using reinforcement learning and contextual bandits (ε-Greedy, UCB, Thompson Sampling).- Implemented off-policy learning techniques (BCT cloning, doubly robust estimators), improving service levels by 8%.- Designed robust evaluation methods using inverse propensity scoring and doubly robust estimators.-----------------------------------------Demand Forecasting – Chemical Industry------------------------------------------ Built ensemble time series models (Naive, Croston, Exp. Smoothing, THETA, ARMA) using Rank-1 holdout blending.- Improved forecast accuracy by 11% (MAPE), unlocking $1.4M/year in potential savings.- Applied ABC-XYZ segmentation to identify candidates for touchless forecasting.------------------------------------------------Inventory Forecasting – Pharmaceutical Industry------------------------------------------------- Benchmarked models across 4/8/12-week horizons using 6MA, 12MA, and univariate Rank-1 models.- Achieved 14% improvement in MAPE using Random Forest and XGBoost for on-hand inventory prediction.- Used SHAP values to explain variable impact and ensure model transparency. Show less

• 

  Mastercard AI Garage

  Nov 2022 - now

  Led end-to-end development and delivery of multiple AI/GenAI solutions, overseeing problem discovery, MVP design, stakeholder engagement, and cross-functional team management (6–7 members):---------------------------------------AI for Dispute & Chargeback Reduction---------------------------------------Built predictive models using Boosting algorithms and Graph Neural Networks to proactively minimize transaction disputes and chargebacks.-------------------------AI in Stand-in Processing-------------------------Designed an AI-driven approval engine that mirrors issuer-specific transaction behaviors to enable smarter decisioning during stand-in scenarios.----------------------------KnowYourTransaction (KYT)----------------------------Developed a GenAI product (deployed on databricks) that explains complex data insights and model decisions in natural language, enhancing transparency and user trust.-------------------------------Mastercard LLM Security Suite-------------------------------Created a comprehensive security suite to assess LLM vulnerabilities by generating adversarial prompts, evaluating model responses, and recommending mitigation strategies. Show less

  • Manager | Lead Data Scientist

    Sept 2023 - now

  • Senior Data Scientist

    Nov 2022 - Sept 2023