Paras Tak

I’m a Machine Learning Engineer who builds Python-based ML systems that are fast, reliable, and production-ready. As a Python Developer Intern at Actisoft IT Services, I refactored legacy code and implemented efficient data structures to achieve a 25% increase in processing efficiency, while automating large-scale data manipulation workflows with Pandas and NumPy.

I’ve delivered impactful AI projects, including NeuralRecon—an NLP RAG vulnerability scanner reaching 87.5% Recall@5 across 105,210 NVD CVEs—using a two-stage BM25 + SecureBERT 2.0 pipeline with FAISS HNSW search and Intel OpenVINO XPU-accelerated inference. I also built a reinforcement learning reasoning agent with 80% task-success using Actor-Critic + curriculum learning, and an end-to-end Streamlit AutoML app for ingestion, preprocessing, and model evaluation; my cybersecurity specialization and EC-Council certifications keep my work grounded in real-world threat research.