Data Science Projects in New DelhiData Science Projects in New DelhiI specialize in transforming complex datasets into intuitive, visually compelling dashboards that drive strategic decision-making. With deep expertise in BI tools and data analytics, I design interactive, real-time reporting systems that help businesses uncover trends, monitor KPIs, and optimize performance. My approach combines data storytelling, clean UI/UX, and scalable architecture to deliver impactful insights.
š¹ My Project Portfolio:
š 1. Executive Business Intelligence Dashboard
Built a dynamic KPI dashboard for C-level executives
Integrated multiple data sources (CRM, ERP, marketing platforms)
Delivered real-time insights on revenue, growth, and performance
Tools: Power BI, SQL, Azure
š 2. E-commerce Analytics Dashboard
Designed a dashboard tracking sales, customer behavior, and conversion rates
Implemented cohort analysis and funnel visualization
Improved marketing ROI tracking by 40%
Tools: Tableau, Python, Google Analytics
š 3. Financial Performance & Forecasting Dashboard
Developed financial models and forecasting dashboards
Visualized cash flow, P&L, and budget vs actuals
Enabled data-driven financial planning
Tools: Power BI, Excel, DAX
š 4. Operations & Supply Chain Dashboard
Created logistics tracking and inventory monitoring system
Real-time alerts for stock levels and delays
Reduced operational inefficiencies
Tools: Looker Studio, BigQuery
š 5. AI-Powered Analytics Dashboard
Integrated machine learning predictions into dashboards
Automated anomaly detection and trend analysis
Delivered predictive insights for decision-making
Tools: Python, TensorFlow, Streamlit Deepfake Detection System
Problem: Deepfake accessibility creates real risksāmisinformation, identity fraud, digital trust erosion. I built a production-grade detector: fast, accurate, explainable, CPU-deployable.
Approach: Transfer learning with ResNet-50 (25.5M params, pre-trained ImageNet-1K V2). Rather than training from scratch on limited data, I leveraged rich hierarchical features that transfer exceptionally well to detecting artifact patterns (blending boundaries, color mismatches, compression artifacts).
Key Decisions:
Face-Centric Preprocessing: OpenCV DNN crops faces to 224Ć224, concentrating attention on artifact-rich regions (eyes, mouth, jawlines).
Custom Classification Head: Bottleneck design (2048ā512ā1) with Focal Loss, label smoothing, progressive unfreezingāprevents overfitting on ~4,000 Ciplab images.
Explainable AI: Custom Grad-CAM generates attention heatmaps showing exactly where the model detected artifacts.
Real-World Robustness: Trained on JPEG compression, Gaussian noise, blur augmentations to handle compressed video feeds.
Results:
AUC-ROC: 0.9424 | Accuracy: 87.25% | Precision/Recall: 87.27%/84.81%
Inference: <100ms per face on CPU
Live on Hugging Face Spaces (no GPU)
Tech: PyTorch, OpenCV, Albumentations, Gradio. Images, videos, live webcamāone codebase.
Demonstrates: End-to-end AI engineering with deliberate trade-offs (efficiency over raw accuracy), training discipline (Focal Loss, Cosine Annealing), and production-first thinking (CPU compatibility, interpretability, real-world robustness).
Demo: https://huggingface.co/spaces/Shri04/deepfake-detector 
