Data Analysis & Insights
- Data Cleaning & Preparation
- Exploratory Data Analysis
- SQL Analysis & Reporting
©2026
DATA Scientist & AI/ML DEVELOPER
About
Hey!
I'm Tejas Saini — an AI/ML engineer who builds things to understand them. I take real-world problems from raw data to working, deployed machine learning products.
My work spans the full pipeline: scraping and cleaning data, feature engineering, model selection and tuning, and serving models through FastAPI APIs people can actually use from the browser.
I'm currently building ML + AI projects and taking on freelance work — from a quick feasibility question to a full end-to-end ML product. If a knot can be untied with data or a model, I'd like to hear about it.
How I work — Learn · Build · Test · Document · Repeat
- 01Learn
- 02Build
- 03Test
- 04Document
- 05Repeat
Currently building ML + AI projects
From idea to launch. Clean, scalable digital products built to move fast, stay simple, and perform in real-world use, driven by clarity, structured systems, and intentional design.
What I do
Services
Machine Learning Solutions
- Predictive Models
- Regression & Classification
- Model Evaluation & Tuning
Data Pipelines & Automation
- Data Wrangling
- Feature Engineering
- Automated Data Workflows
AI / ML-Powered Applications
- AI & ML Prototypes
- Model Integration
- Interactive Web Applications
Featured projects
Selected work.
01 — Portfolio
AutoValue — Indian Car Resale Price Predictor
Know what your car is really worth.
A machine learning project that predicts used-car resale prices from vehicle characteristics, with a focus on data preprocessing, feature selection, model evaluation, and deployment.
Python · Pandas · Scikit-learn · FastAPI
R² 0.32 · MAE ₹298,758 · Gradient Boosting
Project Details
- Problem
- It's hard for a buyer to tell whether a used-car resale price is fair. AutoValue predicts a fair resale price from the vehicle's characteristics.
- Data
- A dataset of used vehicles with their recorded attributes, cleaned and prepared into model-ready features.
- Approach
- A regression pipeline built around preprocessing, feature selection, and model evaluation — then picking the model that performed best on held-out data.
- Deployment
- The model is served through a FastAPI API and reachable from the browser via the live demo.
- Results
- Final model is a tuned Gradient Boosting Regressor with an R² of 0.32, a Mean Absolute Error of ₹298,757.94 and a Root Mean Squared Error of ₹364,899.35 — tuned with a learning rate of 0.2, 50 estimators, and a max depth of 3.
02 — Portfolio
RentDilli
Know your Delhi rent. List with confidence.
A machine-learning-powered web application that estimates the monthly rental price of residential properties in Delhi based on property characteristics and location information.
Python · Pandas · Scikit-learn · XGBoost · FastAPI · JavaScript
R² 0.74 · MAE ₹7,755 · XGBoost
Project Details
- Problem
- Tenants and landlords in Delhi lack a reliable, data-driven way to estimate fair monthly rent, leading to uncertainty and inconsistent pricing across the market.
- Data
- A scraped dataset of residential rental listings across Delhi, cleaned and prepared into model-ready features including property size, location, and amenity information.
- Approach
- A regression pipeline comparing multiple models including XGBoost, with feature engineering and hyperparameter tuning — then picking the best-performing model on held-out data.
- Deployment
- The trained model is served through a FastAPI backend and consumed by a JavaScript frontend, accessible via the live demo.
- Results
- Final model is a tuned XGBoost Regressor with an R² of 0.74, a Mean Absolute Error of ₹7,755.43 and a Root Mean Squared Error of ₹20,827.55 — tuned with a learning rate of 0.2, 50 estimators, and a max depth of 7.
Toolbox
Tools I work with.
- Python
- NumPy
- Pandas
- Matplotlib
- Scikit-learn
- OpenCV
- MediaPipe
- FastAPI
- Git
- GitHub
- XGBoost
- SQL
- MySQL
- Seaborn
- SQLAlchemy
- Jupyter
- VS Code
Reading
Reading & references.
- 01Hands-On Machine Learning with Scikit-Learn, Keras & TensorFlow[ ML ]
- 02Practical Statistics for Data Scientists[ Stats ]
- 03Essential Math for Data Science[ Math ]
Resume
My resume.
Tejas Saini
Data Scientist · AI/ML Engineer · Freelancer
tejassaini.ds@gmail.com github.com/tejassainids linkedin.com/in/tejas-saini
Summary
Data scientist and machine learning engineer who builds things to understand them. I take problems from real-world data to deployed ML products — scraping and cleaning data, feature engineering, model selection and tuning, and serving models behind APIs and browser front-ends. Available for freelance and contract work.
Skills & Tools
ProgrammingPython · SQL · JavaScript
DatabasesMySQL
LibrariesNumPy · Pandas · Matplotlib · Seaborn · SQLAlchemy
Machine LearningScikit-learn · XGBoost · OpenCV · MediaPipe · Feature Engineering · Hyperparameter Tuning
Tools & DeploymentFastAPI · Jupyter Notebook · VS Code · Git · GitHub
Core SkillsData Cleaning · EDA · Data Visualization · Dashboard Development
Featured Projects
AutoValue — Indian Car Resale Price Predictor
Python · Pandas · Scikit-learn · FastAPI
Regression pipeline that predicts fair used-car resale prices. Tuned Gradient Boosting model (R² 0.32 · MAE ₹298,757.94 · RMSE ₹364,899.35) served through a FastAPI API with a live browser demo.
RentDilli — Delhi Rental Price Estimator
Python · Pandas · Scikit-learn · XGBoost · FastAPI · JavaScript
ML-powered web app estimating monthly rent for Delhi properties. Scraped dataset, feature engineering, and a tuned XGBoost model (R² 0.74 · MAE ₹7,755.43 · RMSE ₹20,827.55) behind a FastAPI backend and JavaScript frontend.
Olist E-Commerce Data Analysis using SQL & Python
Python · SQL · MySQL · Pandas · NumPy · Matplotlib · Seaborn
- Designed a MySQL database and imported 100k+ e-commerce records using SQLAlchemy.
- Analysed business questions with joins, aggregations, subqueries, and CASE statements.
- Conducted EDA and built a business dashboard covering revenue, customers, sellers, reviews, and deliveries.
- Published a documented GitHub project with SQL scripts, notebooks, and visualizations.
Currently Building
ML + AI Projects · LLMs / GenAI
Applying the learn — build — test — document — repeat cycle to new ML and LLM product ideas.
Method
Learn
Build
Test
Document
Repeat
References
Available on request.