Full-Stack AI Engineer
Engineering end-to-end artificial intelligence pipelines and scalable full-stack software systems.
// Get to know me
About Me
B.Tech CSE-AI/ML student with hands-on experience engineering end-to-end artificial intelligence pipelines and scalable full-stack software systems.
Passionate about applying supervised learning, probabilistic modeling, and data-driven software architecture to solve complex real-world problems.
Eager to deepen expertise under industry leaders and contribute to impactful AI and software solutions.
// What I work with
Technical Skills
Programming
Web & DB
AI/ML Tools
Data
Core Concepts
// What I've built
Featured Projects
Military Aircraft Detection & Multi-Object Tracking (MOT)
An end-to-end computer vision and MLOps pipeline designed to detect, classify, and continuously track 102 distinct classes of military aircraft in high-speed, dynamic video feeds. Trained on a highly imbalanced dataset of 4,600+ images (25.9M parameters), the model achieved elite metrics on a strictly isolated validation set: Precision (Box P): 84.7%, mAP50: 79.4%, Classification Loss: 1.041.
Features & Architecture
Engineered a persistent JSON-based splitting mechanism for zero data leakage. Built a resilient cloud MLOps pipeline on Kaggle (Tesla T4 GPUs) that auto-saves optimizer state to bypass 12-hour session timeouts. Integrated ByteTrack with custom IoU thresholds for high-speed tracking without ID switching.
Limitations (PoC)
Detecting military aircraft presents a unique CV challenge due to fine-grained visual similarities, such as distinguishing highly similar delta-wing fighters at various altitudes and angles.
Indian Domestic Flight Fare Predictor
A Machine Learning web application that predicts domestic flight prices in India based on historical pricing data. Reverse-engineered airline dynamic pricing logic using an XGBoost Regressor, achieving an R² score of ~97% on historical testing data (July-Aug 2023).
Real-Time Cognitive Fatigue and Drowsiness Monitor
Engineered an end-to-end CV pipeline using OpenCV and Haar Cascades. Trained a MobileNetV2 CNN using TensorFlow/Keras, achieving >94% classification accuracy for alert vs. fatigued states optimized for low-latency CPU.
Hostel/PG Management SaaS Platform
Built a responsive full-stack platform using React, Tailwind CSS, and a Python/Flask backend. Architected a horizontally scalable MySQL database capable of processing millions of records without latency.
// My journey
Education & Certifications
Vellore Institute of Technology-AP
B.Tech in CSE-AI/ML
Certifications
Data Structures and Algorithms (DSA)