Tanish Sanghavi · Data Science & Software Engineering B.Tech (ICT-CS) student at DAIICT. I turn complex data into clear insights through machine learning pipelines and interactive business dashboards.
I learn best by building practical, hands-on projects. When working with data, I follow three core principles:
- I. Focus on Real-World Impact: Pure accuracy is often a misleading metric. I optimize my models for real-world consequences. For example, in healthcare, missing a sick patient is dangerous. That is why I manually tuned my ML model's threshold to 35% to maximize Recall and minimize false negatives.
- II. Clean Data First: A model or dashboard is only as good as the data behind it. I never build a solution without first doing thorough Exploratory Data Analysis (EDA), handling missing values, and structuring the data properly.
- III. Built for the User: Whether it is a machine learning app or a BI report, the end-user experience matters. I build clean web applications and interactive dashboards so non-technical users can actually understand and act on the data.
| Project & Stack | Architecture & Impact |
|---|---|
| Clinical Heart Disease Predictive Tracker ↳ Python · Scikit-Learn · FastAPI · Vercel |
Built an end-to-end machine learning pipeline from scratch. Instead of relying on default settings, I used a custom 35% safety threshold to hit 97% Recall, prioritizing patient safety. Deployed as a web app using a decoupled frontend and backend architecture. |
| Food Express — Enterprise BI Analytics ↳ Power BI · DAX · Power Query · ETL |
Designed an interactive Power BI dashboard to analyze over 154K orders and 20K customers. Created custom DAX calculations to track late-delivery risks and measure repeat-customer revenue ($5.08M), helping guide business strategy. |
| Regional Retail Sales Data Model ↳ SQL · Star Schema · Excel |
Built a solid Star Schema data model (1 Fact table + 4 Dimension tables). Wrote over 25 time-intelligence calculations to track retail performance, running sums, and sales trends over time. |
| Fake Product Review Detection System ↳ Python · Scikit-Learn · Gradient Boosting · SHAP · FastAPI · React |
Engineered an optimized machine learning pipeline deploying a Gradient Boosting model that achieved a 93.3% ROC-AUC and 86.6% F1-Score. Multi-signal architecture synthesizes NLP metrics (lexical diversity) and behavioral metadata flags to isolate bot patterns. Enforced a strict chronological temporal data split to eliminate look-ahead bias and integrated SHAP for explainable visual audit trails. |
(Click the project titles to view the complete repository, methodologies, and source code).
[ Languages ] · Python · SQL · DAX · C · C++
[ Machine Learning & Data Science ] · Scikit-Learn · Pandas · NumPy · Predictive Modeling · Statistical Analysis · Hyperparameter Tuning
[ Business Intelligence & Data Viz ] · Power BI · Matplotlib · Seaborn · ETL Pipelines · Star Schema Design · Data Cleaning
[ Tools & Deployment ] · Git & GitHub · FastAPI · PostgreSQL · Supabase · Vercel · Render · Jupyter Notebook
- Data Visualization with Power BI — Microsoft & Great Learning (April 2024)
I am currently pursuing my B.Tech in ICT-CS at DAIICT, Gandhinagar (Batch of 2028, CGPA: 8.03).
While I continue to sharpen my skills in traditional Machine Learning and BI, my immediate focus is expanding into GenAI by learning about Retrieval-Augmented Generation (RAG) and Agentic AI workflows.
📍 Based in Ahmedabad, Gujarat, India.
- Email: tanishsanghavi2@gmail.com
- LinkedIn: linkedin.com/in/tanish-sanghavi-a44b873b6