I am deeply passionate about advancing the frontiers of Artificial Intelligence and Machine Learning through rigorous research and practical implementation. My work spans multiple domains including:

• 🧠 Deep Reinforcement Learning - Exploring novel architectures and training methodologies
• 🔒 Adversarial Machine Learning - Investigating model robustness and security vulnerabilities
• 🤖 Neural Network Optimization - Developing efficient architectures for resource-constrained environments
• 📊 AI for Finance - Applying ML techniques to quantitative trading and risk management
• 🏗️ Systems Engineering - Building scalable AI infrastructure and distributed systems

Primary Areas:

• Adversarial attacks on deep neural networks (FGSM, PGD, C&W)
• Reinforcement learning in complex environments
• Federated learning and privacy-preserving ML
• Neural architecture search and automated ML
• AI safety and interpretability

Methodologies:

• PyTorch-based deep learning implementations
• Mathematical optimization and algorithmic design
• Empirical evaluation and statistical analysis
• Reproducible research practices

Investigating FGSM attacks on MNIST classification

• Research Question: How do small perturbations affect model confidence?
• Methodology: Implemented FGSM adversarial attacks with varying epsilon values
• Key Finding: 89.71% accuracy drop with ε=0.3, revealing critical vulnerabilities
• Impact: Contributes to understanding of neural network robustness

Building RL agents without high-level libraries

• Approach: Custom MLP implementation for Deep Q-Learning
• Environments: CartPole, Lunar Lander, Taxi
• Contribution: Educational framework for understanding RL fundamentals

Designing proof-of-work consensus mechanisms

• Architecture: P2P network with socket programming
• Features: Mining algorithms, digital signatures, longest chain consensus
• Future Work: CUDA acceleration for mining optimization

Short-term Objectives:

• Advancing adversarial robustness research
• Exploring federated learning applications
• Contributing to open-source ML libraries

Long-term Vision:

• Pursuing graduate research in AI safety
• Publishing in top-tier ML conferences
• Building production-ready AI systems

I'm actively seeking research collaborations in:

• Adversarial Machine Learning
• Reinforcement Learning Applications
• AI Safety and Interpretability
• Efficient Neural Architectures

Open to: Research internships, paper collaborations, and technical discussions with fellow researchers.

For Research Inquiries:

• 📧 Email: sharmaachintya49@gmail.com
• 💼 LinkedIn: achintyasharma47
• 📝 Medium: @sharmaachintya49