"Engineering isn't always perfect. It's about how you approach the failures and iterate towards the solution."
I am navigating the convergence of Machine Learning, IoT Architecture, and Software Engineering at Woxsen University. For me, AI isn't an isolated lab experiment—it's a system to be engineered, monitored, and scaled.
Every node is a concept. Every edge is a connection. I don't build toy models; I construct functional pipelines.
| 🛡️ Operational Pillars | Details |
|---|---|
| 💡 Hardware Invention | Granted an IoT Connectivity Device Design Patent by the Gov. of India (No: 470097-001) for originating authentic machine-to-machine communication hardware. This involved complete PCB structuring, signaling architecture, and power-efficient micro-controller communication layers. |
| 🧠 Algorithmic Integrity | Re-engineered a fragile 99% accuracy model down to an unbreakable 92% real-world ready system using deep K-Fold CV, aggressive layer pruning, and L2 Regularization. I don't chase validation metrics—I chase deployment robustness. |
| ☁️ Continuous Deployment | Focused on dragging robust ML models out of Jupyter Notebooks and fusing them into functional SaaS web infrastructure. I integrate React front-ends with Flask/FastAPI backends connected to clustered inference nodes. |
My journey is backed by an intense curriculum of advanced development, rigorous machine learning math, networking, and modern stack operations. Here are my verified credentials:
| Certification Level | Course & Focus | Issuing Institution |
|---|---|---|
 Advanced Algorithms |
Java Programming: Solving Problems with Software | Duke University (via Coursera) |
| Data Engineering |
Data Structures Using Python | Packt (via Coursera) |
| Theoretical Compute |
Quantum Computing For Everyone | Fractal Analytics (via Coursera) |
| Software Architecture |
Object-Oriented Design | University of Alberta (via Coursera) |
| IoT & Embedded C |
The Arduino Platform & C Programming | UC Irvine (via Coursera) |
| Microcontrollers |
Interfacing with the Arduino | UC Irvine (via Coursera) |
| Network Defense |
Introduction to Cyber Attacks | NYU Tandon School of Engineering |
| Deep Hardware |
Nanotechnology and Nanosensors | Technion - Israel Institute of Tech |
| Frontend Expertise |
Frontend Developer (React) | HackerRank |
| Backend Logistics |
Go (Basic) | HackerRank |
| Professional Skills |
Write Professional Emails in English | Georgia Institute of Technology |
What makes my work unique is the crossover between Deep Tech AI, Hardware Operations, and Consumer Web Platforms. I approach every individual project as an enterprise deployment.
🧠 StrokeGuard.ai — Clinical Predictive Modeling
Overview:
This isn't just a simple .predict() script. It actively analyzes disparate, noisy patient health vectors (age, glucose levels, BMI, work status) to output precise, statistically-backed stroke risk classifications.
Technical Execution:
- 🎯 Algorithmic Focus: Developed logic for highly imbalanced medical datasets utilizing SMOTE oversampling and rigorous threshold calibration techniques.
- 🛡️ Optimization: Obsessively pushed to minimize false negatives, an absolute necessity in real-world clinical triage operations where errors mean lives.
- ⚙️ Deployment Architecture: Abstracted complex pipeline states into a modular inference framework, bridging notebook experimentation with actionable dashboards.
🌾 AgriSathi — Agri-Tech Decision Engine (V3 Evolution)
Overview:
A multi-parameter agricultural intelligence system. It pulls historical soil N-P-K data, current climatic markers, and regional pH values to dynamically output actionable farming strategies.
Technical Execution:
- 🧠 Inference API: A core Flask backend seamlessly serves predictive intelligence based on real-time and simulated metrics.
- ⚛️ Frontend V3 Architecture: Built a completely responsive
.js/React visualization layer. Fixed critical UI routing, and integrated a persistent browser-based shopping cart backend via local storage. - 🚀 Simulated Infrastructure: Upgraded the system to a production-ready Web App on GitHub Pages, ensuring every interaction triggers persistent real state changes rather than static visual effects.
🏥 MedStack Platform — Enterprise Hospital Management
Overview:
Built the raw core scaffolding for end-to-end medical administration, eliminating asynchronous latency across heavily networked hospital wings.
Technical Execution:
- ⚡ Backend Pooling: Designed highly efficient backend pooling structures for concurrent doctor scheduling tasks and dynamic patient billing pipeline calculations without deadlocking.
- 📡 Live Telemetry: Engineered a WebSocket-driven infrastructure to handle real-time bed telemetry and critical updates simultaneously across multiple integrated client surfaces.
🎯 AI Computer Vision Deployments
Overview:
A dual-focus portfolio pushing serverless deployment limits, containing both a high-fidelity Age & Gender Detection model and an ML Object Detection inference unit.
Technical Execution:
- ☁️ WebAssembly Integrations: Deployed serverless python logic across Hugging Face and GitHub Pages via Gradio Lite, specifically pinning version conflicts to load large
huggingface-hubCaffe models purely client-side. - 👁️ Model Integrity: Architected complex image manipulation streams ensuring latency-free inferences using heavy multi-classification parameters directly within the browser sandbox.
🖲️ Patented M2M Hardware Node (IoT)
Overview:
Researched, diagrammed, structured, and patented an IoT Connectivity Device that robustly bridges local edge sensor-data continuously with global cloud ingestion architectures.
Technical Execution:
- 🏆 Legal Standing: Officially Granted an Innovation Patent (Design Patent No: 470097-001 by Intellectual Property India).
- 🔋 Embedded Operations: Reduced packet loss overhead using proprietary C-level micro-controller transmission paths, aggressively optimizing voltage power-draw constraints for remote edge computing arrays.
💞 LifeLink — Organ Donation Ecosystem
Overview:
An interactive full-stack coordination directory aimed at tracking user commitments, points, and gamified donation milestones.
Technical Execution:
- 🗄️ Environment Hardening: Rescued production from persistent HTTP 500 errors by aggressively altering backend database schemas with targeted dynamic table updates across InfinityFree servers.
- 👤 Functional UI Integration: Upgraded previously broken dashboard profiles with real working API linkages for fully tracking donations within the operational MySQL relational pipeline.
🔥 Expand to view specific loadouts
