I build end-to-end machine learning systems focused on prediction, automation, APIs, and real-world problem solving using Python, Flask, and Scikit-learn.
🎓 BSc in Data Science and Programming, IIT Madras (CGPA: 8.4)
🤖 Passionate about Machine Learning, Data Science, and Applied AI
📊 Skilled in Machine Learning, Data Analysis, SQL, Backend APIs, and Predictive Modeling
⚡ Experienced in building end-to-end ML systems, backend services, and data-driven applications
🤝 Strong collaborative experience through Agile-based software development projects
💼 Actively seeking opportunities in Machine Learning Engineering, Data Science, and Applied AI
- 🤖 Machine Learning using Python — Simplilearn
- 📊 Power BI for Beginners — Simplilearn
- ☁️ Introduction to Cloud Computing — Simplilearn
- 🟢 LeetCode: 44+ Problems Solved
- 🟢 HackerRank: SQL (Advanced)
End-to-end ML system built to identify high-risk customers using classification models and ensemble learning.
- Processed 7K+ customer records with 40+ features
- Built Pipeline-based preprocessing workflows
- Applied Stratified Cross Validation and threshold optimization
- Achieved ROC-AUC ~0.85
- Developed Ensemble Voting Classifier
- Deployed real-time prediction API using Flask
Python • Scikit-learn • Flask • Pandas • NumPy
🔗 Repository: Customer Churn Prediction System
Full-stack healthcare support platform built using Flask, Vue.js, and SQLite to assist elderly users and caregivers through reminders, emergency assistance, and accessible user interfaces.
- Led Primary User Home Page development
- Developed reminder, emergency alert, and help-status APIs
- Integrated Geolocation API for SOS location sharing
- Contributed to Text-to-Speech reminder functionality
- Wrote pytest test cases for Primary User APIs
- Participated in frontend-backend integration and system testing
- Developed 15+ REST APIs across multiple user roles
- Implemented JWT-based authentication and role-based access control
- Built medication reminder and emergency alert systems
- Enabled live geolocation sharing during SOS events
- Designed normalized relational database schema using SQLite
- Supported Primary, Secondary (Caregiver), and Tertiary (Healthcare/NGO) user workflows
- Achieved high API reliability through comprehensive pytest-based testing
Flask • Flask-RESTful • Vue.js • SQLite • JWT • Pytest • JavaScript Geolocation API • Web Speech API • Bootstrap 5
🔗 Repository: CareConnect
Academic team project completed as part of IIT Madras Industry 4.0 (GAABS 4.0).
Built a machine learning-based prioritization framework to identify high-risk misinformation content and improve the effectiveness of limited human fact-checking resources.
- Combined TF-IDF text features with engagement and metadata signals
- Applied Random Forest classification for misinformation detection
- Designed an Expected Harm scoring framework for review prioritization
- Achieved ~1.87× moderation efficiency improvement
- Intercepted ~56% of harmful engagement while reviewing only the top 30% of prioritized content
- Collaborated throughout project development, evaluation, and final presentation
- Participated in exploratory analysis, result interpretation, and validation activities
- Contributed to project documentation, reporting, and presentation preparation
Python • NLP • TF-IDF • Scikit-learn • Random Forest • Pandas
🔗 Repository: Risk-Based Misinformation Review System
Multi-class classification system using XGBoost for crime category prediction.
- Achieved 95.1% accuracy
- Applied feature engineering and scaling
- Used K-Fold Cross Validation
- Performed hyperparameter tuning
- Achieved Leaderboard Rank 222
Python • XGBoost • Scikit-learn • Pandas
🔗 Repository: Crime Category Forecasting
- ML System Design
- FastAPI for ML Deployment
- LLM Applications
- Docker & MLOps Fundamentals
- Advanced SQL & Query Optimization
- 💼 LinkedIn: https://www.linkedin.com/in/geetha-akula
- 📧 Email: geethaakula123@gmail.com
⭐ I enjoy building scalable machine learning systems that combine data science, backend engineering, APIs, and real-world business impact.