
I am a dedicated AI professional with a focus on scalable automation and generative intelligence. My expertise lies in designing agentic workflows that transform complex data into actionable insights. Having authored multiple research papers in blockchain and federated learning, I bring a rigorous analytical approach to every deployment.
Architecting intelligent systems through rigorous engineering. A specialized focus on Large Language Models (LLMs), RAG architectures, and autonomous agentic workflows.
Care Guide
Designed and deployed HIPAA-compliant Healthcare AI systems processing sensitive patient data with zero leakage.
Developed RAG-based conversational AI chatbots, improving enterprise query resolution accuracy by 60%.
Built autonomous AI agentic workflows for clinical documentation automation, reducing repetitive overhead by 50%.
Code24 PTY LTD
Engineered automated NDIS platforms to streamline participant onboarding and workforce rostering via custom API integrations.
Built multilingual conversational AI chatbots using IBM Watsonx Assistant for global client outreach.
Designed voice-first AI assistants integrating n8n, Retail AI, and CRM systems for real-time lead qualification.
United International University
Led laboratory sessions for Data Structures & Algorithms and AI-focused coding intensives.
Mentored 150+ students in Python-based machine learning fundamentals and complex problem-solving strategies.
Guided undergraduate research teams in debugging large-scale neural network architectures and dataset curation.
Engineering intelligent systems that bridge the gap between raw data and actionable insights through LLMs, custom RAG frameworks, and Agentic workflows.
AI-powered business consultation platform designed for contract analysis, legal document reviews, and strategy support.
FastAPI backend and React frontend integrated with LangGraph.
LLM conversational interface with JWT and 4k context window.
Multilingual RAG pipeline supporting Bengali and English search queries, optimized for e-commerce search.
Hybrid semantic retrieval using intent classification.
Persistent vector storage across 12+ categories.
Natural language to SQL assistant querying relational databases using Model Context Protocol (MCP).
Integrated LLaMA-3.1-8b with Streamlit UI query executor.
Interactive syntax-highlighted real-time execution.
AI-driven workflow tool converting natural language instructions into CSV analytics operations.
Integrated Gemini 1.5 Flash for automated visualization.
Chat-based analytics for real-time reports and charts.
10+
Projects Completed
04
Core Stacks
98%
Model Accuracy
1+
Year Experience
Engineered solutions at the intersection of Federated Learning, Privacy-Preserving AI, and Medical Diagnosis. Focusing on efficiency metrics and scalable architectures.
Proposed a privacy-preserving blockchain-based medical image sharing framework integrating subject sensitive hashing (SSH) and IPFS storage, achieving 98% data integrity accuracy across multiple healthcare imaging datasets.
Developed a federated learning-based ensemble deep learning model with explainable HiRes-CAM and SHAP attribution for decentralized lung cancer detection, achieving 98.26% accuracy.
Designed an attention-based CNN architecture (ARS-CNNSA) for classification of arsenic-induced skin conditions, outperforming standard models with 91% accuracy and 90% F1-score.
Introduced a clustering-driven classification framework leveraging unsupervised instance similarity to reduce labeling dependency and enhance performance of decision tree and ensemble-based classifiers.
Developed an ensemble-based phishing detection system combining Random Forest, Logistic Regression, and Gradient Boosting to improve generalization and detection accuracy in cybersecurity applications.

Bsc in Computer Science & Engineering
GPA
3.67 / 4.00
Graduated
OCT 2024
United International University
IBM TechZone
HackerRank Certified