Building the impossible at 3 AM. Breaking systems to make them better.

I'm a first-principles thinker who architects systems, not just writes code. My passion lies in:

• 🧠 AI/ML Research: Custom neural architectures, multi-modal learning, pushing computational boundaries
• ⚡ Rapid Prototyping: Full-stack apps from concept to deployment in under 60 minutes
• 🔧 Production Systems: Enterprise-scale applications handling 5000+ units in production
• 🔬 Innovation: 2+ years researching alternative computational paradigms beyond binary logic

"I don't memorize syntax. I discover mechanisms."

The Problem: Google Maps' official place_tracker sample was deprecated and crashing on modern Android devices, blocking developers worldwide.

My Solution:

• ✅ Migrated entire codebase from Dart 2 to Dart 3 with full null safety
• ✅ Upgraded Android Embedding from deprecated v1 APIs to v2, eliminating runtime crashes
• ✅ Tested on modern devices ensuring compatibility across Android versions
• ✅ Documented breaking changes for seamless migration

Status: 🟢 CLA Signed | Pull Request Submitted to Google Engineering

📌 View Pull Request →

Impact: Unblocked hundreds of developers relying on Google's official sample code.

Architecting production-ready applications with focus on scalability and speed

• Built real-time sales tracking visualization tool for business intelligence
• Integrated AI-driven predictive insights for sales trend forecasting
• Achieved sub-second query response times on large datasets
• Tech Stack: React, Node.js, Python, TensorFlow, Jax

• Developed full-stack restaurant management system streamlining order workflows
• Optimized for high-performance UX with sub-second latency
• Reduced order processing time by 40% through efficient architecture
• Tech Stack: Flutter, Firebase, REST APIs

• Developed custom CNN architecture for automated MRI diagnostics
• Trained neural network from scratch achieving high diagnostic accuracy
• Implemented data augmentation pipeline for medical imaging datasets
• Tech: Python, TensorFlow, Computer Vision

• Research into synchronized multi-modal training (audio, video, text)
• Custom neural architecture studying pattern resonance across modalities
• Novel wavelength-based data representation mechanisms
• Tech: Python, NumPy, Signal Processing

• Architected system managing 5,000+ fire extinguisher units for TCS deployment
• Built offline-first mobile app with 100% data integrity in zero-connectivity zones
• Reduced audit reporting time by 60% with real-time web dashboard
• Tech: Flutter, Dart, Firebase, QR Technology

• Engineered hardware-software IoT system for real-time soil NPK analysis
• Programmed PSoC 5LP microcontroller with custom sensor drivers
• Optimized for 24+ hour battery runtime for field deployment
• Tech: PSoC, Embedded C, Python, UART, IoT Sensors

• 🔬 Research: Exploring alternative computational paradigms beyond binary logic
• 🧠 AI Optimization: Running trillion-parameter models on consumer hardware
• 🌐 Open Source: Contributing to Flutter and Google's ecosystem

class Sharvesh:
    def __init__(self):
        self.mindset = "First Principles"
        self.coding_hours = "3 AM"
        self.approach = "Build Systems, Not Code"
        
    def solve_problem(self, problem):
        # I don't follow books. I discover mechanisms.
        solution = self.architect_from_scratch(problem)
        return solution
        
    def handle_pressure(self):
        # Pressure doesn't stop me. Mundane work does.
        return self.code_something_impossible()