Apr 2024 – Aug 2025 · 1 yr 4 mos · Mumbai Metropolitan Region, India

• A Proptech platform with AI first principles solving problems and providing seamless robust capabilities for real estate and adjacent industries

May 2023 – Apr 2024 · 11 mos · On-site

• Successfully initiated and nurtured the application of data science methodologies within Bajaj's Personal Loan Cross-Sell team
• Directly accountable to Business Heads for managing and mentoring a dynamic team of 12 professionals
• Devised strategic approaches to amplify the existing customer base, effectively balancing portfolio risks
• Coordinated with multiple data providers, including CRIF, CIBIL, EXPERIAN, Vahan, and MSEB, ensuring comprehensive data integration
• Significantly enhanced lead generation by integrating Large Language Models (LLM) such as GPT-2, LLAMA, and FLAN-T5
• Pioneered in developing agent training programs and managed to handle times through innovative zero and few-shot learning techniques
• Spearheaded prompt engineering and fine-tuning of models, achieving optimal performance
• Evaluated LLMs using HELM, Rouge, and MMLU metrics and applied PEFT and LoRa methodologies to streamline model space
• Elevated the likelihood of customers opting for personal loans from 0.4% to a targeted 0.8%
• Conducted an in-depth analysis of market segments, identifying growth opportunities in underrepresented areas
• Set and pursued the ambitious goal of establishing 20 new potential branches to expand Bajaj's 7% market share in personal loans
• Effectively controlled credit risk, maintaining key Risk KPIs such as FEMI, CBB, 0+6MOB, and 3MOB
• Enhanced amount loss management and collection efficiency, contributing to a robust risk management framework
• Identified and addressed customer segments prone to lodging complaints with TRAI regulators, reducing promotional complaints
• Successfully leveraged various Bureau on-us and off-us features, inquiries, EPFO data, bank statements, and social features to develop and implement multiple approaches and models
• Achieved recognition and approval from PLCS regional Business Heads for the developed models, with ongoing performance monitoring and evaluation

Aug 2018 – Dec 2022 · 4 yrs 4 mos

• Addressed the challenge of automating the matching of handwritten checks, including the dollar amount and signature verification
• Developed and trained Convolutional Neural Networks (CNN) and Bidirectional Recurrent Neural Networks (RNN) models to create a sophisticated in-house Optical Character Recognition system
• This system was designed to accurately interpret human handwriting on payment and deposit checks, significantly reducing manual effort and minimizing human errors
• Result: Successfully achieved an automated system that accurately processes approximately 80% of payment and deposit checks
• Tackled the substantial financial penalties (approx. $100MM/year) faced by banks due to regulatory non-compliance.
• Utilized Natural Language Processing (NLP) techniques to analyse text transcripts, effectively enforcing regulatory compliance
• The project involved monitoring over 40 regulations in call centres, including the Telephone Consumer Protection Act (TCPA)
• Analytical Solution: Implemented intelligent call monitoring using advanced text analytics and deep learning architectures such as BERT, Skip-gram, Transfer Learning, CBOW with n-gram approach, Bidirectional-LSTM, Logistic, and tree-based models
• Result: This approach led to a remarkable saving of $6MM monthly, the implementation of 3 new compliance policies, the elimination of manual call listening, and the removal of third-party vendors
• Focused on enhancing efficiency by removing Level 1 (L1) support and maximizing Interactive Voice Response (IVR) containment
• Applied predictive analytics for caller intent prediction and understanding of call complexity
• Implemented capacity planning for call centres using time series analytics and topic modelling
• Result: Developed an enterprise-level text analytics tool using Shiny (R), and a comprehensive Customer Journey Analytical tool, encompassing nearly all text mining capabilities

Oct 2017 – Aug 2018 · 10 mos

• Enhancement of Hotel Website Conversion:
• Spearheaded implementation of hotel discounting strategies within a defined budget, employing genetic programming for optimization
• Developed and fine-tuned a web personalization recommendation system, focusing on up-selling and cross-selling
• Utilized tree-based models for identifying key features and the Apriori algorithm for mining sequences with high confidence and support.
• Conducted comprehensive A/B testing on various propensity models
• Result: Boosted the hotel website's monthly conversion rate from 6% to 10%, generating approximately 20-25 lakh in additional revenue. Also achieved a 6% conversion rate in the churned customer segment
• Optimization of Upselling and Cross-Selling in Cab Services:
• Led the analysis and utilization of tree-based models to identify actionable features for the cab business
• Implemented similarity-based approaches for identifying and targeting new customer segments with specialized campaigns
• Result: Successfully increased upselling and cross-selling by 18% in a quarter, contributing to an additional 10-15 lakh in revenue per month.
• Strategic Use of Time Series Analysis in Flight Fare Management:
• Analysed flight fare variations over time to create a sense of urgency for online ticket bookings
• Executed customer profiling to identify and capitalize on market opportunities

Nov 2016 – Oct 2017 · 11 mos

• BPOST Mail Classification System:
• Developed an end-to-end solution for categorizing BPOST's mail, including digitally scanned items
• Result: Enhanced mail sorting efficiency and accuracy
• ExxonMobil Fraud Detection System:
• Implemented a credit card fraud detection system for ExxonMobil
• Result: Improved fraud detection, reducing financial risk
• Adidas Sales Forecasting:
• Led sales forecasting for Adidas' new apparel and shoe launch
• Result: Enabled targeted regional sales strategies and inventory management

Nov 2015 – Nov 2016 · 1 yr

• Led the development and implementation of machine learning models in retail and telecom sectors
• Proficient in big data management and analysis using Apache Spark and Python tools, enhanced data processing and insight generation
• Utilized key algorithms for regression, clustering, time series, and classification to improve pattern recognition and strategic data utilization

Sep 2014 – Nov 2015 · 1 yr 2 mos

• Crafted analytical solutions for customer targeting, market segmentation, and product design at Vodafone Germany and Italy, enhancing marketing and product strategies
• Conducted demand forecasting and brand valuation, improving market trend prediction and brand value assessment
• Executed profit and price analyses, leading to more effective financial planning and pricing strategies
• Developed fraud detection methods in the telecom sector, notably reducing fraud incidents and enhancing customer trust

Dec 2010 – Aug 2014 · 3 yrs 8 mos

• Served as Senior Product SME at Amdocs, handling end-to-end aspects of telecom rating and billing products (Ensemble and Enabler), enhancing operational efficiency
• Supported IBM's statistical team with data cleaning and transformation, ensuring data accuracy and usability
• Developed data pipelines for the analytics team, streamlining data processing and analysis workflows