MBA (General Management), IIM Nagpur, 2025–27 · B.E., BITS Pilani

I'm an analyst at heart. On every project, the report was never where I wanted to stop: the next step was always to automate it, so the model keeps learning from new data and the insight reaches the person who has to act on it. This profile is where I build those next steps.

How I work with AI: Claude, ChatGPT and Gemini are part of my daily work for analysis, drafting and code. I use agentic analyst tools for exploratory work, and I test AI output before trusting it: every project here shows how.

A cancellation-risk model turned into a running system: scores new bookings weekly, monitors its own accuracy and drift, retrains itself, and passes each week to an LLM analyst whose every number is checked against the model output. Python scikit-learn Streamlit Claude / Gemini API prompt design

Re-sized pre-approved loan offers on what each customer has proved they can repay, modulated by a LightGBM risk score chosen on holdout, not cross-validation. Scores 89.6/100 on the brief's metric against 60.0 for the incumbent logic, on the same book. Python LightGBM credit risk policy design

Team project on a 6-table e-commerce dataset. Predictive models, statistical tests, and a cost-based decision model for reducing returns, including an honest near-random model result that we tested against chance instead of over-claiming. Python scikit-learn statistics decision modelling

Three years of real invoice data from a consumer-durables dealership. Two reconciliation catches turned a "collapse and recovery" story into steady erosion, and showed the only growing category is the least cross-sold. Python market basket RFM reproducible analysis

Benchmarked 6 peers across 302 facilities; projected headcount and return-to-office demand for 9 cities × 9 business lines to 2029; built a consolidation cost model. Case study only; company data stays private. forecasting benchmarking Excel modelling

Linked P&L, balance sheet and cash-flow budget built from volume and realisation drivers, with bear/base/bull scenarios and a break-even test on input costs. FP&A financial modelling Excel

Four live-project modules: brand-survey visualisation (11,109 responses), sales forecasting, CRM/NPS analysis and cash-flow analysis. Excel forecasting survey analytics

Python (pandas, scikit-learn, LightGBM, Streamlit) · Excel (Solver, advanced modelling) · Power BI · Tableau · LLM APIs (Claude, Gemini) · prompt engineering