Md. Jahidul Islam

• Architected Iris, a real-time multimodal voice-and-vision AI studio deployed at iris-puce.vercel.app using Next.js, TypeScript, and Ollama (LLaVA, llama3.1).
• Built Mosaic RAG, a multimodal retrieval engine achieving 76.7% hybrid MRR on SciFact benchmarks vs. 73.5% BM25 baseline.
• Integrated Web Speech API, local OSS models, and Vercel production pipelines for live AI demos.
• Published open-source repos with documented benchmarks and deployment guides on GitHub.

• Designed and deployed a production software marketplace serving users at vibeflow.tech.
• Built React frontend, Node.js API, product catalog management, and Vercel CI/CD pipeline.
• Delivered maintainable full-stack architecture with modern UX and deploy-ready infrastructure.

• Led a team delivering a full-stack project-management platform with auth, workspaces, and Kanban boards.
• Merged three sprint deliverables into a unified PHP/MySQL application with shared session architecture.
• Coordinated frontend practices, code reviews, and final integration across Task 1–3 modules.

• Delivered responsive, performance-optimized business websites for multiple clients.
• Implemented accessible front-end patterns and production-ready deployment workflows on Vercel.
• Built custom React utilities including an identifier-length validator for cross-language schema safety.

• Published 50+ production Android applications on Google Play across business, games, and utilities.
• Built Kotlin apps with privacy-first, offline-first architecture for booking, real estate, wholesale, and legal services.
• Managed Play Store releases, client branding, and production maintenance for international business clients.

• Unified ML/CV benchmark platform — 20+ models across 4 tracks with reproducible CLI and live leaderboards.
• CNN achieves ~90–92% on Fashion-MNIST vs ~82–85% flatten+RF; YOLOv8s reaches 65.8% mAP@50 on Hard Hat detection.

• Real-time multimodal AI studio combining voice input and vision analysis with local OSS model support.

• Multimodal RAG evaluation harness with published SciFact hybrid-retrieval benchmarks (76.7% MRR).

• Full-stack software marketplace with catalog management and live production deployment.

• AI-powered code review tool with multi-language support, delivering structured feedback in seconds.

• Disaster relief management platform with resource tracking and role-based coordination workflows.

American International University-Bangladesh (AIUB)

• Coursework in algorithms, data structures, software engineering, and relational database systems (Oracle SQL, MySQL, PostgreSQL).
• Applied full-stack web development, AI/ML foundations (Python, PyTorch, TensorFlow), and cloud tooling (AWS, Docker, Vercel) across academic and project work.
• Project lead, Group 08 — delivered a full-stack PHP/MySQL Web Technologies capstone with authentication, workspaces, and Kanban boards.