Data Scientist & Analytics Engineer | Machine Learning · Deep Learning · Advanced SQL · Power BI

I build predictive machine learning systems, statistical models, and high-performance data analytics architectures. My work spans deep learning architectures from scratch, end-to-end production ML pipelines (FastAPI, MLflow, threshold tuning), advanced SQL cohort models, and dimensional data modeling with Star Schema and DAX.

Currently pursuing a Minor in Artificial Intelligence & Data Science at IIT Mandi, with an active focus on scalable machine learning systems and Agentic AI workflows.

Tech Stack: Python, LightGBM, XGBoost, MLflow, FastAPI, Pydantic, SHAP

• Developed an end-to-end ML pipeline with custom class-imbalance strategies (SMOTE, cost-sensitive weighting, undersampling).
• Optimized classification decision thresholds directly against financial cost matrices to maximize net fraud recovery.
• Built explainability modules using SHAP and tracked iterative model runs via MLflow.
• Deployed a production-ready real-time scoring service with a validated FastAPI schema and automated unit tests.

🔗 Repository: github.com/kra10m/fraud-detection (or subfolder in your ML repo)

Tech Stack: Python, NumPy, PyTorch, Matplotlib

• Implemented vectorized multi-layer Fully Connected Neural Networks (FCNN) from scratch using pure NumPy.
• Conducted diagnostic research on gradient propagation, vanishing gradients, and learned spatial representations.
• Benchmarked architectures across MNIST and Tiny ImageNet with PyTorch, running systematic ablation studies on depth and initialization.

🔗 Repository: github.com/kra10m/fully-connected-neural-networks (or subfolder in your ML repo)

Tech Stack: PostgreSQL, Advanced SQL, Window Functions

• Processed and analyzed 100,000+ retail transaction records using high-performance analytical SQL queries.
• Built dynamic behavioral customer segments to evaluate retention curves and lifetime values.
• Engineered cohort-based revenue tracking pipelines utilizing window operations (LAG, LEAD, moving averages).

🔗 Repository: github.com/kra10m/SQL_Projects_Data_Analytics

Tech Stack: Microsoft Excel, Power Query (M), Power Pivot, DAX, Star Schema

• Engineered an enterprise-grade reporting solution eliminating row-level formula bloat via the xVelocity/VertiPaq in-memory engine.
• Designed a normalized Star Schema data model (1:* cardinality) linking fact tables across customer, product, and dynamic calendar dimensions.
• Developed defensive DAX measures for Year-over-Year (YoY) revenue changes, profit distributions, and basket sizes (AOV).

🔗 Repository: github.com/kra10m/Ecommerce_Excel_Portfolio

For broader exploratory work, specialized queries, and modular components, explore these dedicated collections:

• Machine Learning & Data Science Portfolio — Deep learning benchmarks, predictive modeling pipelines, and production APIs.
• SQL Analytics Repository — Complex queries, window functions, and database schema architectures.
• Python Data Analytics Repository — Exploratory data analysis, scraping pipelines, and statistical analyses.
• Power BI Business Intelligence — Multi-source dashboards, DAX measures, and stakeholder KPIs.

June 2024 – December 2024

• Built Power BI dashboards and automated reporting workflows for educational operations and revenue tracking.
• Analyzed student enrollment, fee collection metrics, and operational datasets to guide administrative choices.
• Monitored KPI pipelines to maintain consistent data hygiene across internal dashboards.

• Minor in Artificial Intelligence & Data Science — IIT Mandi (2025 – Present)
• Bachelor of Computer Applications (BCA) — LN Mishra Institute (2021 – 2024)

• 📧 Email: atandon027@gmail.com
• 💼 LinkedIn: linkedin.com/in/abhinav-tandon-212a54388
• 🐙 GitHub: github.com/kra10m