• 🔐 Building real-time anomaly detection, malware scoring & UEBA systems at C-DOT, processing telemetry from 10,000+ endpoints
• 🤖 Built Stock Agent — an end-to-end agentic AI system for real-time stock analysis with dual-LLM orchestration (Claude API + Ollama) and RAG
• 🧠 Research in Federated Learning Security, Adversarial Attacks & Phishing Detection
• 📚 Published in ICONIP, IJCNN & SINCOF on federated learning resilience and phishing defense
• 🌱 Working with LangChain, LangGraph, MCP, and RAG for production agentic AI applications

End-to-end agentic AI system for real-time stock market analysis — live NSE data, news retrieval, and personal-document RAG (29,000+ chunks) via a dual-LLM (Claude API + Ollama) conversational agent, deployed full-stack with FastAPI + React.

Studying cognitive biases in phishing detection using ML/DL + GloVe embeddings. (Published: ICONIP 2022)

X-vector system using MFCC feature extraction (96% training accuracy), sponsored by the National Language Translation Mission. Live-demo GUI built with PyTorch and Pygame.

Flask app for real-time bird species prediction using audio classification.

Analysis of separable vs non-separable datasets (image & tabular) with ML experiments.

• ICONIP 2022 — Cognitive Biases in Phishing Emails & Human Decision-Making
• SINCOF 2023 — BATFL: Battling Backdoor Attacks in Federated Learning
• SINCOF 2023 — RAFT: Evaluating Federated Learning Resilience Against Threats
• IJCNN 2023 — On Robustness of Split Neural Networks Against Data Poisoning Attacks

• Agentic AI with LangChain & LangGraph — IBM (Coursera)
• Build RAG Applications: Get Started — IBM (Coursera)
• Develop Generative AI Applications: Get Started — IBM (Coursera)
• Neural Networks & Deep Learning — DeepLearning.AI
• Linear Algebra for ML & Data Science — DeepLearning.AI

• 💼 LinkedIn
• 🧑‍💻 GitHub
• 📚 Google Scholar
• ✉️ Email: mayankkumar1303@gmail.com

⭐️ "Exploring the intersection of AI and Security to build robust, intelligent systems."