I build backend systems for voice, automation, and cloud infrastructure. For the past three-plus years at VoiceCradle I've designed and shipped the systems behind an AI-powered customer communication platform: Amazon Connect IVR call flows, Amazon Lex and LLM-backed chatbot layers, Lambda and DynamoDB services, and the real-time dashboards the team uses to run it.

Before that, at USAHomeBid, I worked on API performance and database optimization for a real-estate platform, using query optimization and caching to significantly cut page load times.

I'm most interested in event-driven architecture, distributed systems, and running AI in production. My stack is Java/Spring Boot on AWS, plus Python and Docker. On the research side, I work on transfer learning for medical image classification.

• 🔭 Currently: AI voice & IVR automation on Amazon Connect · cloud-native backend systems
• ☁️ AWS Certified: Solutions Architect – Associate, Cloud Practitioner
• 📊 Research interests: transfer learning, medical image classification, data leakage in ML evaluation, speech processing
• 🐳 DevOps: Docker, Nginx, Linux, CI/CD, observability (OpenSearch / ELK)
• 🎓 Education: B.Sc. in Computer Science & Engineering, BUBT (expected Dec 2026)

• AWS Certified Solutions Architect – Associate (SAA-C03), Aug 2026 · Verify on Credly
• AWS Certified Cloud Practitioner (CLF-C02), Jul 2026 · Verify on Credly
• Odoo 19 Certified Functional · View certificate

1. Analyzing Regional Variations in Bangla Speech: A Voice Classification using MFCCs International Conference on Artificial Intelligence for Sustainable Engineering and Innovation (ICAISEI), 2026 · ResearchGate District-level Bangla voice classification (Dhaka vs. Chittagong) from MFCC features, comparing classical ML models.