Sagnik Bhanja +91-9178594317
Indian Institute of Technology, Kharagpur
Education
Degree Institute Board / University CGPA/Percentage Year
B.Tech Indian Institute of Technology, Kharagpur IIT 8.52 2020-2024
Senior Secondary AISSCE CBSE 94.40% 2020
Matriculation AISSE CBSE 96.2% 2018
Professional Experience
• HDFC Bank Ltd July ’24 - Present
Methodology /approach to ordinal regression in credit risk modelling
– Worked as credit risk analyst in the Credit Risk Analysis and Innovation department of HDFC Bank’s data science division; applica-
tions of ordinal regression models to expedite and enhance credit scoring & other customer-quality metrics prediction were explored
– Used machine learning libraries to apply proportional odds model and continuation ratio model for predicting loan recovery rate and
credit score respectively; numerical and categorical features were chosen by EDA involving various feature engineering techniques
– Built customized evaluation techniques to verify the model assumptions on the lines of Brant-Wald hypothesis test and ,further, used
recall values, KS-Statistics and AUC and ROC curves, to evaluate the prepared model
Internship
• HDFC Bank Ltd May ’23 - July ’23
Feature importance of gradient boosting models and application of reject inferencing
– Created modules that allow for feature importance computation of trained gradient boosting models on non-training dataset i.e Out
of time, Out of sample, and, Post development datasets was done for a multitude of popular boosting algorithms
– Increased the reliability of feature stability evaluation for ML pipelines using XGBoost and LightGBM by avoiding re-fitting on
non-training datasets; also explored source code to understand tree building, boosters, and the influence of hyperparameters
– Wrote functions to incorporate the reject inferencing techniques namely, Fuzzy augmentation and Parceling, into scorecard pipelines
Projects
• Quantitative Risk Analysis | IIM Ahmedabad Aug ’23 - Sept ’23
Objective: To develop and backtest VaR models for accurate risk estimation across multiple asset classes
– Assessed risk for NIFTY 50, USD/INR, & GOLD by developing VaR models; focussed on historical VaR estimation technique
– Implemented Historical, GARCH, Parametric & Monte Carlo VaR models, utilizing 1000-day rolling backtesting on 20-year data
– Achieved lowest exceedance rates, with Historical VaR - 0.0094% for NIFTY 50, 0.0133% for USD/INR, and 0.0078% for GOLD
• Model Validation | IIT Kharagpur Nov ’23 - Dec ’23
Objective: To Validate FX Pricing models and explore stochastic volatility models
– Developed an internal FX library with numerical techniques as MC, Multi Level MC, Batched MC and Quasi RNG as sobol sequences
– Performed stress, convergence testing, and sensitivity analysis for exotic FX products like FVA, Variance Swaps, Volatility Swaps
– Implementing sophisticated stochastic volatility models, as Dupire and Heston, to improve price and risk assesment for FX products
• Bachelor’s Thesis Project | Dept of Humanities and Social Sciences, IIT Kharagpur Oct ’23 - Jan ’24
Objective: To assess the efficacy of financial inclusion schemes and suggest improvements to existing inclusion index measures
– Developed novel financial inclusion indices incorporating demand-side metrics for efficient classification
– Used Principal Component Analysis to build a financial inclusion index in two stages; latent variables were modelled intermediately
– Used the model on time series data to estimate the effect of financial inclusion on GDP. ADF and PP tests analyzed stationarity,
ARDL measured co-integration, and ECM investigated short-term relationships between variables.
– Conducted a bibliometric analysis of research conducted in the field of financial inclusion in India and applied the same analysis
techniques to Chinese counterparts of Indian research papers; Utilized keyword co-occurrence and author co-citation methods
Technical Skills
+
• Programming Languages : C, C++, Python, SQL • Tools: NumPy, Pandas, Matplotlib, Pytorch, Tensorflow, Azure Databricks
Relevant Coursework
Probability and Statistics | Real Analysis | Transform Calculus | Partial Differential Equation | Numerical Analysis | Machine
Learning Foundations & Applications | High Performance Computing | Programming and Data Structures | Economics
Achievements
– Was among top 5% of the batch at the end of first year and secured a change of department consequently
– Was among 800 students in the country to be awarded the NTSE scholarship in 2018 upon clearing a two-stage selection process
– Qualified to appear for INMO(Indian National Mathematics Olympiad) - 2017 upon clearing preliminary stages at the state level
– Qualified to appear for INPhO(Indian National Physics Olympiad) - 2016 upon clearing preliminary stages at the state level
