Your Friendly Neighborhood AI Engineer & UX/UI Designer — training neural nets, drafting high-fidelity mockups, and crafting premium, user-centered digital products.
Three panels. One origin. Infinite possibilities.
It started with a fascination — a late-night encounter with deep learning algorithms and predictive databases. Studying Artificial Intelligence & Data Science at B.S. Abdur Rahman Crescent Institute of Science & Technology changed everything. I was immediately bitten by the AI & UI/UX bug.
Like any hero, I trained relentlessly — learning Python, TensorFlow, PyTorch, and training Convolutional Neural Networks (CNNs). I conquered data preprocessing and engineered bulletproof voice authentication models and deep-learning image distenders in intense, fast-paced hackathons.
Realizing that powerful algorithms need beautiful interfaces, I emerged as a designer. Today, I build high-fidelity interactive wireframes, custom user flows, and mobile layouts in Figma. Bridging raw mathematical logic with seamless, stunning, human-centered product designs.
Each domain represents a skill I've mastered in battle.
SHIELD-approved project dossiers. Eyes only.
Multi-service home assistance mobile app. Connects homeowners seamlessly with certified professionals for cleaning, electrical repair, plumbing, and wellness support.
Service professional counterpart application. Empowers local service agents with booking pipelines, scheduling managers, digital invoice generators, and real-time earnings charts.
AI diet planner translating predictive biometric inputs into personalized nutrition schedules. Built on neural network algorithms for customizable macronutrient distribution models.
Unified campus companion hub. Synchronizes university lecture notes, event calendar updates, automated peer study alerts, and interactive event trackers via real-time Firebase endpoints.
Fintech budgeting companion. Employs real-time charts and automated ledger controls to trigger expense predictions and manage financial limits securely in high-performance structures.
Dual-engine audio emotion ecosystem. Features a 90% accurate TF-IDF bigram NLP model to classify lyrics, paired with a GPU-accelerated Flask web app ("The Divination Chamber") powered by RoBERTa transformers and a magical interactive wizard UI.
Robust, passwordless voice authentication system. Employs a CNN model trained on MFCC voice features with an ensemble verification mechanism (combining CNN, average cosine similarity, and DTW distance) to prevent spoofing and replay attacks.
Lightweight CNN-based image sharpening model trained via self-supervised knowledge distillation using sharp ground truth images as a fake teacher. Optimized using MSE, MS-SSIM, and edge-aware Sobel losses on the GoPro Dataset.
Every job was a boss fight. Here's my win record.
Defeated the "Biometrics Security Threat" by architecting and training a secure Convolutional Neural Network (CNN) Voice Authentication Model on 500+ voiceprints. Honored with the prestigious "Next Potential Idea Award" for merging cutting-edge deep learning with an intuitive, human-centered UI verification flow.
Conquered the "Image Distortion Monster" by developing a state-of-the-art StudentCNN model for high-resolution image sharpening. Successfully applied advanced **Knowledge Distillation** parameters and designed Multi-Scale Structural Similarity (MS-SSIM) loss logic to compress teacher model accuracy with ultra-low latency.
Defeated "The Budgeting Leakage" by establishing the backend architecture for Loot Locks, a personal UPI financial-control application. Engineered real-time Firebase syncing, secure REST API integrations, and robust user authorization checks to prevent overspending on active active budget limits.
Messages from citizens of the web. No filter, no edits.
Shine the Spider-Signal. I will respond within 24 hours or the city burns.