data.json

{
  "experience": [
    {
      "logo_light": "./assets/capital-one-light.svg",
      "logo_dark": "./assets/capital-one-dark.svg",
      "alt": "Capital One Logo",
      "company": "Capital One",
      "company_abbreviation": "CO",
      "title": "Senior Data Analyst",
      "dates": "May 2025 - Present",
      "duties": [
        "Senior Data Analyst within the Refinance Line of Business (ReFi LOB)."
      ]
    },
    {
      "logo_light": "./assets/temple-light.png",
      "logo_dark": "./assets/temple-dark.png",
      "alt": "Temple University Logo",
      "title": "Data Analyst",
      "dates": "Jan 2024 - May 2025",
      "duties": [
        "Automated weekly lead reports using Excel VBA, saving 8+ hours/month and enhancing responsiveness to data issues.",
        "Created Tableau dashboards for enrollment funnel and campaign performance, delivering student engagement insights that improved conversion by 12%.",
        "Conducted Python-based EDA and A/B testing, increasing conversion by 12% and engagement by 20%.",
        "Structured data with SQL and pandas, enabling root cause analysis and metric development for internal dashboards.",
        "Delivered insight decks to university leadership, improving planning decisions with clear visualizations."
      ]
    },
    {
      "logo_light": "./assets/unilever-light.png",
      "logo_dark": "./assets/unilever-dark.png",
      "alt": "Unilever Logo",
      "title": "Product Analytics Consultant",
      "dates": "Jul 2024 - Dec 2024",
      "duties": [
        "Built Power BI dashboards for pacing and inventory risk, enabling $3M+ in savings across product lines.",
        "Automated ingestion of 15M+ records using Databricks pipelines for anomaly detection and real-time reporting.",
        "Built forecasting models using regression/trend methods, boosting demand accuracy to 96% and reducing stockouts.",
        "Defined cross-functional KPIs with product, finance, and supply teams, standardizing P&L and POS data tracking.",
        "Delivered weekly metrics decks and executive dashboards to guide decision-making on replenishment and performance."
      ]
    },
    {
      "logo_light": "./assets/groupM-light.png",
      "logo_dark": "./assets/groupM-dark.png",
      "alt": "GroupM Logo",
      "title": "Marketing Data Analyst",
      "dates": "Jan 2022 - Jun 2023",
      "duties": [
        "Managed SQL-driven performance marketing attribution models for $3M+ media spend, improving ROAS by 22% and informing cross-channel strategy.",
        "Designed CLV segmentation models and pacing dashboards to reduce churn by 18% and increasing recurring purchases.",
        "Created forecasting tools and automated Excel dashboards for client media planning and spend strategy.",
        "Uncovered 15% untapped markets via stakeholder interviews/demographics, guiding a $1.2M campaign for Colgate-Palmolive."
      ]
    },
    {
      "logo_light": "./assets/first-economy-light.png",
      "logo_dark": "./assets/first-economy-dark.png",
      "alt": "First Economy Logo",
      "title": "Media Operations Analyst",
      "dates": "Mar 2021 - Jan 2022",
      "duties": [
        "Streamlined reporting automation using Excel macros and Google Data Studio, reducing reporting lag by 70%.",
        "Led weekly QA checks, discrepancy resolution, and report refresh processes for campaign operations.",
        "Delivered dashboards and insights to internal teams and clients, informing logistics of media delivery."
      ]
    }
  ],
  "projects": [
    {
      "image": "./assets/bellion.svg",
      "title": "Bellion: Zero-Trust Financial AI Agent",
      "description": "Everyone talks about using AI or Agentic AI, but nobody explains how to apply it to real-world problems. This project bridges that gap. Bellion is a headless, zero-trust financial architecture built on OpenClaw, orchestrating automated daily briefings, proactive budget auditing, and secure WhatsApp integrations\u2014all running locally to ensure absolute data privacy.",
      "github": "https://github.com/Nishant-Iyer/OpenClaw-Bellion"
    },
    {
      "image": "./assets/credit-card-fraud.png",
      "title": "Credit Card Fraud Detection",
      "description": "Built an end-to-end production-grade MLOps pipeline and interactive Streamlit explainability dashboard to detect credit card fraud under extreme class imbalance (0.17% positive rate). Features robust preprocessing, a PyTorch-based Autoencoder for unsupervised anomaly feature representation, and a Stacking Ensemble (XGBoost + LightGBM + Logistic Regression) calibrated against a financial business utility function (reducing total fraud loss overhead by 75.8%). Integrates local SHAP explanation waterfalls and Autoencoder reconstruction breakdown tools.",
      "github": "https://github.com/Nishant-Iyer/Credit-Card-Fraud-Detection.git",
      "liveDemo": "https://credit-card-fraud-detection-rvtygaztzdva8ecohigihs.streamlit.app"
    },
    {
      "image": "./assets/distribution-optimization.png",
      "title": "Global Distribution Centers Optimization",
      "description": "Architected an enterprise-grade spatial optimization platform using K-Medoids (geodesic PAM), K-Means (Euclidean 3D), MILP (capacitated PuLP/CBC), and PyTorch Spherical Gradient Descent on the S^2 unit sphere. Incorporates World Bank Logistics Performance Index (LPI) infrastructure readiness penalties, dynamic purchasing power demand weighting, and scenario-based forced-DC placements, visualized on an interactive 3D Web Globe.",
      "github": "https://github.com/Nishant-Iyer/Global-Distribution-Optimization-Center.git",
      "liveDemo": "https://global-distribution-optimization-center-wudamnacmixpl2i56vb36j.streamlit.app"
    },
    {
      "image": "./assets/nfl-analysis.png",
      "title": "NFL Gambling Market Analysis",
      "description": "Built an end-to-end MLOps pipeline and interactive Streamlit dashboard to predict NFL Over/Under betting market inefficiencies. Features custom feature engineering (Elo ratings, travel distances, weather impact), walk-forward chronological backtesting, and live Kelly Criterion sizing with dynamic SHAP local prediction explanations.",
      "github": "https://github.com/Nishant-Iyer/NFL-Gambling-Market-Analysis.git",
      "liveDemo": "https://nfl-gambling-market-analysis-nrtofnwx4xqxvpqj7mcvlj.streamlit.app"
    },
    {
      "image": "./assets/wellbore-prediction.png",
      "title": "Wellbore Geology Prediction (Kaggle)",
      "description": "Engineered a geophysically-constrained Viterbi Dynamic Programming path finder and 2D neighbor dipping plane regression solver to predict True Vertical Thickness (TVT) in horizontal wellbores. Bypasses missing logs via automated typewell entry depth back-calculation, achieving a validation RMSE of 11.08 ft on unseen wells and a perfect 0.0000 RMSE score on the public leaderboard via train-match override.",
      "github": "https://github.com/Nishant-Iyer/Rogii-Wellbore-Geology-Prediction.git",
      "liveDemo": "https://rogii-wellbore-geology-prediction-u9ejzh9ghvs8cdmnftsajc.streamlit.app"
    }
  ]
}