Deepthi Makkapati

About

Software Development Engineer with experience building enterprise backend systems using Java, Spring Boot, and PostgreSQL. Focused on scalable REST APIs, secure authentication and authorization, and AI-powered workflow automation.

Experience

• Software Development Engineer (AI)
  AIRA HR – NR Consulting · Hyderabad, Telangana, India
  Jan 2026 – Present

  Joined as a Software Engineer Intern in Jan 2026 and converted to full-time Software Development Engineer (AI) in Jul 2026. Built backend systems for enterprise education and AI-recruitment platforms using Java, Spring Boot, and PostgreSQL, including multi-schema data models with JPA/Hibernate mappings and JSONB fields. Designed a multi-role RBAC system synced with Keycloak group management, integrated JWT authentication with internal user resolution, enforced semester-based data isolation, optimized PostgreSQL analytics queries, fixed assessment engine issues, built webhook persistence for interview scoring callbacks, automated recruiter workflows with Playwright, and built/maintained an n8n-powered AI sales pipeline.

Education

• Integrated B.Tech + M.Tech
  Jawaharlal Nehru Technological University Hyderabad
  Computer Science & Engineering · 2021 – 2026

Skills

Tools / apps / platforms

Projects

• Face Recognition Attendance System
  Python, OpenCV, Haar Cascade, LBPH, Flask

  Built an automated attendance system using OpenCV, Haar Cascade Classifier, and LBPH for real-time facial detection and recognition, with secure logging/reporting via a Flask-based interface for enrollment and monitoring.

• EatRight – Intelligent Diet Planning System
  Python, Flask, HTML, CSS, Bootstrap, Jinja2

  Developed a full-stack Flask application generating personalized diet plans from health metrics, lifestyle, and fitness goals. Integrated an AI-assisted chatbot for dietary recommendations alongside user authentication and persistent meal-planning data management.

• Explainable AI for Financial Fraud Detection & Credit Scoring
  Python, InterpretML, SHAP, LIME, Scikit-learn, Pandas

  Built an interpretable ML pipeline for credit risk prediction using Explainable Boosting Machines (EBM), SHAP, and LIME, prioritizing transparency over black-box accuracy. Designed a modular preprocessing and training pipeline with feature-level explainability and interactive visualizations for trustworthy credit scoring.