Headline

Solving BFSI @Sarvam

About

As a Data Scientist at HDFC Bank, I specialize in leveraging advanced analytics, machine learning, and AI-driven solutions to solve complex business challenges. My expertise spans Generative AI, time series forecasting, and predictive modeling, with a strong focus on innovation and operational efficiency. Throughout my career, I have contributed to multiple industries, from banking and finance to logistics and retail. At HDFC Bank, I have been instrumental in integrating Generative AI solutions across departments, enhancing ID verification through facial recognition models, and improving chatbot efficiency with Retrieval Augmented Generation (RAG). My work on AI-driven document clustering has streamlined HR processes and document management. Previously, at Stylumia, I developed machine learning-based forecasting models to optimize store inventory and SKU planning, improving business performance. During my tenure at RAAHO, I pioneered the data science team, leading groundbreaking projects in freight pricing, demand-supply optimization, and supplier recommendations, significantly increasing efficiency and profitability. With a strong academic foundation in Statistics and Data Science from NMIMS and IIT Madras, I am passionate about applying data-driven insights to drive strategic decision-making. I have also been recognized in national and international forums for my work in data visualization and statistical modeling. I am always eager to explore emerging technologies, collaborate on challenging projects, and contribute to the advancement of AI and data science in impactful ways. Let’s connect to discuss innovations in AI, analytics, and data-driven decision-making!

Location

Bengaluru

AI stack

langfuse

langgraph

DSPy

Google ADK

llama.cpp

PydanticAI

Work experience

Applied AI at Sarvam, Mar 2026 - Present · 6 mos, Bengaluru, Karnataka, India, Solving complex problems through Agents in BFSI and Strategic sector

Manager - Data Science at HDFC Bank, Jul 2024 - Feb 2026 · 1 yr 8 mos, Mumbai, Maharashtra, India, Online Reputation Management (ORM) – GenAI Automation Built a GenAI-driven ORM system using domain-tuned LLMs and agentic workflows to automate complaint triaging and response generation. Achieved ~70% workflow automation, ~90% accuracy, reduced resolution time by ~20%, and lowered operational costs by 40–50%. Agentic Document Summarization & Validation Developed an agentic AI-based document summarization and validation platform using LangGraph and fine-tuned LLMs, enabling automated field validation and policy-level reasoning. Reduced manual review effort by 87%, cutting turnaround time from ~2 hours to ~15 minutes. Real-Time ML Models for NBA / INBA (Personal Loans) Designed real-time and near real-time ML models using XGBoost and trigger-based behavioral features to power the bank’s NBA/INBA framework. Delivered a 10× uplift in cross-sell conversions over BAU by improving STP customer identification and offer timing. GenAI-Powered HR Onboarding Q&A Assistant Led a GenAI-based RAG assistant trained on 150+ internal HR documents to automate employee onboarding queries. Improved response accuracy and turnaround time while saving 100+ hours per month through contextual, self-service query resolution.

Data Scientist at HDFC Bank, Jan 2024 - Jul 2024 · 7 mos, Cross-Departmental AI Integration: Played a key role in developing Generative AI products, spearheading the exploration and integration of AI capabilities across various departments to drive innovation and enhance operational efficiency. Advanced ID Verification Algorithm: Developed a cutting-edge ID verification system utilizing MTCNN for precise face detection and FaceNet512 for robust face recognition and image verification. This state-of-the-art algorithm significantly improved security and user verification processes. Chatbot Enhancement with RAG: Implemented a Retrieval Augmented Generation (RAG) model using open-source Large Language Models (LLMs) integrated with Langchain. This enhancement tailored our internal chatbots to provide more accurate and contextually relevant responses, enhancing user experience and support efficiency. AI-Driven Document Clustering: Leveraged Sentence-BERT (S-BERT) for clustering job descriptions, effectively identifying and consolidating documents with similar content. This application reduced redundancy in job descriptions, streamlining HR processes and improving document management efficiency.

Data Scientist (Forecasting and Prediction) at Stylumia, Oct 2023 - Dec 2023 · 3 mos, Hybrid, Strategic Impact and Technical Expertise: Data-Driven Brand Strategy: Utilized advanced data analysis techniques to address key brand challenges, including optimizing store inventory management, strategizing SKU (Stock Keeping Unit) planning, and identifying high-revenue products. This approach enabled the detection of emerging market trends, significantly enhancing overall business performance and strategic decision-making. Advanced Forecasting Model Development: Leveraged Nixtla, a state-of-the-art time series forecasting tool, to develop a hierarchical machine learning-based model. This model was specifically tailored to forecast SKU counts at a store level, facilitating precise inventory management and planning. This initiative not only improved operational efficiency but also supported dynamic market responsiveness.

Data Scientist at RAAHO, Jan 2023 - Aug 2023 · 8 mos, Gurugram, Haryana, India, Pioneered and led the establishment of the data science team, driving innovation and delivering impactful solutions. Created GAMER forecasting model, reducing supplier's time for receiving the freight charges by 98%. Developed data-driven demand and supply zones using clustering techniques with a focus on high profitability. Developed dynamic hybrid pricing model (LSTM+XG-Boost), increasing bid bounds coverage from 15% to 97.5%. Accurately predicted prices hourly, resembling a stock market approach. Created vehicle, and supplier-to-lane recommendations using NCF (Neural Collaborative supplier recommendation, and shipper recommendation systems to optimize the freight marketplace and maximize fill rate. Conducted extensive research in the Online Freight Exchange (OFEX) domain, leveraging insights to design predictive models for real-time inventory management and precise freight charge prediction.

Data Scientist at RAAHO, Jun 2022 - Dec 2022 · 7 mos, Gurugram, Haryana, India, Pioneered and led the establishment of the data science team, driving innovation and delivering impactful solutions. I have actively contributed in building a next-generation "freight marketplace" often referred to as the "Uber for trucks." My involvement in Raaho has allowed me to make significant contributions in the following areas: Designed and Implemented a Supplier Recommendation System:- Utilizing machine learning techniques and economic models, I successfully designed and implemented a supplier recommendation system. This innovative system aimed to match the most suitable suppliers with the specific requirements of users, enhancing the efficiency and effectiveness of the freight marketplace. Developed a Vehicle Recommendation System: I played a pivotal role in developing an indent recommendation system that employed priority analysis. This system revolutionized the process of recommending appropriate vehicles based on their availability and analysis of various vehicle types. Conducted Thorough Research and Summarized Scientific Papers: This allowed me to stay up to date with the latest advancements in the field and provided valuable insights for improving the freight marketplace. My research skills and ability to synthesize complex information helped drive evidence-based decision-making and ensure the product's continuous evolution. By highlighting these key contributions and skills, I aim to showcase my expertise and demonstrate the value I can bring to future endeavors in the freight and logistics industry.

Education

Master's degree, Statistics and Data Science, SVKM's Narsee Monjee Institute of Management Studies (NMIMS)

Bachelor's degree, Data science and Programming, Indian Institute of Technology, Madras