Welcome to my GitHub! I’m a researcher and engineer working at the intersection of Machine Learning, Deep Learning, and Autonomous Agents.
I enjoy exploring problems from first principles, implementing them rigorously, and sharing my work openly here.

• Implement ML & DL algorithms from scratch to understand them deeply.
• Explore reinforcement learning and agentic systems, applying them to real-world challenges.
• Work on generalization, optimization, and interpretability in deep learning models.
• Build open, reproducible projects that others can learn from and build upon.

• Reinforcement Learning & Autonomous Agents
• Neuromorphic & Energy-Efficient AI
• LLMs, MoE, and Agentic Systems
• Applied AI in Healthcare, Agriculture, and Cybersecurity
• AI Safety, Generalization, and Interpretability

• GenAI Research Agent: An autonomous research assistant that uses LLMs to read papers, summarize findings, and suggest new directions for exploration. Repository Link
• Neuromorphic Decision Transformer (arXiv Paper): A spiking neural variant of Decision Transformers, designed to reduce inference energy while maintaining strong policy performance. Repository Link
• SmartTraffic-RL: A reinforcement learning framework for adaptive urban traffic signal control, including a custom gym-style RL environment with SUMO integration. Repository Link
• Repo Ingestion Agent: An agent that ingests GitHub repositories, parses code/documentation, and structures them for downstream LLM tasks. Repository Link
• KibouChan: A chatbot focused on mental wellness and positive interactions, blending conversational flow control with LLMs. Repository Link
• AgroSense (arXiv Paper): A crop recommendation system leveraging soil and climate data, built for precision agriculture and sustainable farming. Repository Link
• AgenticCyberOps: An autonomous cybersecurity pipeline for penetration testing, threat detection, and automated response. Repository Link
• NumPyLiteC: A lightweight C-based reimplementation of NumPy-like operations, designed for speed and minimal footprint. Repository Link
• PDF2Podcast: Converts research papers and PDFs into podcasts using summarization and text-to-speech. Repository Link
• pandas.c: A systems-level reimplementation of core Pandas features in C, exploring performance and efficiency trade-offs. Repository Link
• Handwritten Recognition (MNIST): A deep learning model for digit recognition on the MNIST dataset. Repository Link
• Wave Function Collapse: Procedural content generation using the WFC algorithm. Repository Link
• Stay Productive: A simple productivity app for tracking tasks, focus sessions, and consistent workflow. Repository Link
• Command-Line Game: A lightweight interactive terminal game showcasing UI/UX in the command line. Repository Link

• Signature Verification (CNN, SNN, CSNN): Verifies signatures using convolutional and spiking neural network architectures. Code Link
• Drowsiness Detection: Detects driver drowsiness using data and image inputs for real-time safety applications. Code Link
• Malware Detection (DT, SVM, LR, NN): A comparative study of classical ML and neural approaches for malware detection. Code Link
• Signature Verification (RNN / C-RNN / B-RNN): Explores recurrent architectures (vanilla, conditional, bidirectional) for signature verification. Code Link
• Real Forge Signature Detection: Distinguishes between forged and genuine signatures using image-based models. Code Link
• Evaluate OpenLLMs for Text Summarization: Benchmarks open-source LLMs on summarization tasks, comparing quality vs inference cost. Code Link
• Diabetes Prediction and Classification (Bachelor’s Project): Predicts diabetes types with feature importance analysis, preprocessing, and model comparison. Code Link

• Spiking Decision Transformer: Neuromorphic adaptation of Decision Transformers. Paper Link
• AgroSense: Precision agriculture system for crop recommendation. Paper Link
• Diabetes Prediction and Its Types: Undergraduate research project exploring classification of diabetes subtypes.

• 💻 50+ repositories covering ML, DL, RL, and agentic systems.
• 🧠 Focused on clarity, reproducibility, and teaching through code.
• 🌱 Always experimenting, learning, and documenting along the way.

• Email: [email protected]
• GitHub: Vishal-sys-code
• Blog & Writings: vayishu.bearblog.dev
• X: its_vayishu
• Google Scholar: vayishu-gscholar

Special thanks to my research collaborator Debasmita Biswas for her contributions to projects including AgroSense and Spiking Decision Transformer.

✨ Thanks for visiting! If you find my work interesting, feel free to ⭐ my repos or reach out, I’m always open to collaboration and research discussions.