KARTHIKEYA AYYAGARI

KARTHIKEYA AYYAGARI


About

AYYAGARI KARTHIKEYA Kakinada, Andhra Pradesh, India | +91 8919258592 Mail: karthikeyaayyagari20070@gmail.com LinkedIn: karthikeya-ayyagari-7a592b3a2 PROFESSIONAL SUMMARY CSE (AI & ML) student with hands-on experience in Python development, backend engineering, AI/ML applications, and data security. Interested in building secure, practical software solutions. EDUCATION B.E. Computer Science and Engineering (AI & ML) SCSVMV University, Kanchipuram | 2024–2028 Current CGPA: 9.01/10 (after 2nd year, 4th semester) TECHNICAL SKILLS * Languages: C, C++, Java, Python * Frontend: HTML5, CSS3, JavaScript * Backend: Flask, FastAPI, REST APIs, Uvicorn * Databases: SQL, SQLite, PostgreSQL * Data Formats: JSON, CSV * AI/ML: NumPy, Pandas, Matplotlib, Scikit-learn * Computer Vision: OpenCV, Pytesseract, EasyOCR, DeepFace * Security: DLP, Tokenization, Vault Encryption, OTP, Device Binding, RBAC * Tools: Git, GitHub, VS Code, Streamlit, Pytest * Core Concepts: OOP, Data Structures, Inheritance, Encapsulation, Polymorphism, Abstraction

Skills
Python
Java
React
Flutter
Android
Gemini
PostgreSQL
Projects

1. S.H.A.D.E. (Synthetic Host for Automated Data Extraction)

A local-first data security system designed to protect sensitive information through pattern recognition, tokenization, and secure data rehydration. Includes clipboard monitoring, device binding, OTP authentication, an encrypted local vault, authorization controls, exposure reporting, and audit logging. Designed to work offline without requiring an internet connection for core functionality.

Programming: Python Backend: FastAPI
Uvicorn Database: SQLite Security: Tokenization
vault encryption
Argon2 PIN protection
OTP authentication
device binding Testing: Pytest Other: Clipboard monitoring
pattern recognition
DLP (Data Loss Prevention)

CoercionSafe Banking

A banking security and risk assessment system designed to identify potentially coerced or unsafe transactions. Evaluates contextual, environmental, behavioural, and approval-related factors to calculate a risk score, classify risk levels, and determine whether manual verification is required. The broader project concept includes safety signals such as panic gestures and multi-voice detection.

Programming: Python Backend: FastAPI
Uvicorn Risk Engine: Custom risk scoring and classification logic Security: Authentication
safety signals
manual verification workflows API: REST APIs

Fraud Detection and AI-Based KYC Verification

An AI-assisted fraud detection application designed to support document verification and identity validation. Project modules include document OCR, tampering detection, signature verification, Aadhaar and PAN fraud detection, AI-based KYC verification, and unusual-pattern analysis. The implementation has involved prototypes and individual modules rather than a confirmed fully deployed production system.

Programming: Python Image Processing: OpenCV
Scikit-image OCR: Pytesseract
EasyOCR Facial Analysis: DeepFace Machine Learning: Scikit-learn
anomaly detection Potential Integration: FastAPI for backend services

BankGuard – Unusual Transaction Pattern Analysis

A web-based transaction analysis application designed to import transaction data, inspect unusual patterns, and present results through visual charts and reports. The interface supports spreadsheet processing and PDF report generation.

Frontend: HTML
CSS
JavaScript Visualisation: Chart.js Data Import: PapaParse
XLSX Reporting: jsPDF Data Format: CSV
JSON
spreadsheet files

Signature Fraud Detection System

A Python-based signature verification system that compares submitted signature images with stored reference samples to identify potential mismatches. Uses image similarity analysis and configurable thresholds to make approval or rejection decisions. In the earlier manual testing, the prototype classified 8 out of 10 test cases.

Programming: Python Image Processing: OpenCV
image similarity analysis Core Logic: Similarity thresholds
approval/rejection rules Architecture: Modular preprocessing
comparison
and output components

Bujji – Personal AI Assistant

A modular personal assistant project designed to operate as a single-user assistant with capabilities spanning AI, automation, browser interaction, communication, security, and privacy. The architecture focuses on modular development, automated testing, and the longer-term goal of controlling a laptop through a connected phone while the computer is running.

Programming: Python Architecture: Modular software design Modules: AI
automation
browser interaction
communication
security
privacy Testing: Pytest Development Tools: Git
GitHub

SCSVMV Faculty Portal

A desktop application designed to manage faculty information and attendance records. Provides separate host/admin and faculty workflows and supports importing attendance data from ESSL X990 biometric attendance CSV files.

Programming: Python Frontend/Desktop UI: Tkinter Data Handling: CSV
Python file handling Core Features: Faculty management
attendance records
role-based access
Education

BE-CSE

Sri Chandrasekharendra Saraswathi Viswa Mahavidyalaya.

Aug 2024 to Aug 2028