AI agents and LLM products
From retrieval and model workflows to tool integrations, evaluation and production-ready interfaces.
Explore AI developmentMahmudAlMuhaimin
I build LLM systems, agent backends and the products around them, from the model call to the screen.
Based in Dhaka, Bangladesh, I work on LLM and RAG applications, AI agent backends, and full-stack AI products for local and remote teams.
Open to AI engineer and full-stack rolesBased in Dhaka, Bangladesh. Email is the fastest way to reach me.
- freelance clients served
- 248+
- REST endpoints in SuprBuild Mesh
- 20+
- sketch classifier test accuracy
- ~92%
- faster report queries at my internship
- ~35%
Experience
2020–Present
Freelance software and digital marketing
Bangladesh
Delivered web development, SEO and go-to-market work for 248+ clients across SaaS, e-commerce and travel. CV-reported client growth reached up to 45x within six months of digitization.
Oct 2022 – Jan 2023
Software developer intern
Square InformatiX Ltd.
Built and unit-tested 6 components for an electronic data management system, made report queries ~35% faster, fixed 25+ defects, and contributed to database design, documentation and requirements gathering.
2019 – 2023
BSc Software Engineering
Yunnan University, China
Thesis: an ML-based movie recommendation system. Relevant study areas included machine learning, data structures and algorithms, software architecture, and software project management.
Technical focus
- AI
- LLM applications, RAG, agents, agentic workflows, embeddings, semantic search, AI orchestration
- Languages
- Python, TypeScript, JavaScript, SQL, C++
- Backend
- Node.js, Express, FastAPI, REST, SQLite, MySQL
- Frontend
- React, Next.js, responsive UI
- ML
- PyTorch, TensorFlow, CNN, LSTM, VAE, computer vision, recommender systems
- Engineering
- Software architecture, testing, Git/GitHub, Linux, API integration, product prototyping
How can I help?
Agent backends
Build task APIs, stateful workflows and integrations that keep agent actions observable and controlled.
Agent backend developmentFull-stack product work
Turn a product brief into a working web application, including the frontend, APIs and data flows.
See selected workHiring for your team?
Review my experience, technical focus and selected work for AI engineering and full-stack roles.
View employer information- LLM applications
- RAG
- agents
- agentic workflows
- embeddings
- semantic search
- AI orchestration
- Python
- TypeScript
- JavaScript
- SQL
- C++
- Node.js
- Express
- FastAPI
- REST
- SQLite
- MySQL
- React
- Next.js
- responsive UI
- PyTorch
- TensorFlow
- CNN
- LSTM
- VAE
- computer vision
- recommender systems
- Software architecture
- testing
- Git/GitHub
- Linux
- API integration
- product prototyping
Selected work
Latest writing
2026-10-15 · 2 min read
Designing a task mesh for AI agents
Design notes from SuprBuild Mesh: how agents publish work, complete it, earn reputation and get paid.
2026-10-15 · 1 min read
A checklist before you ship an LLM feature
Seven practical questions I ask before an LLM feature reaches real users.
2026-10-15 · 1 min read
How the text assistant on this site works
How the portfolio's text assistant sends questions to a rate-limited API and falls back to profile-based answers.
2026-10-15 · 2 min read
What is agentic AI? A practical guide to useful, bounded agents
A practical explanation of agentic AI, when an AI agent is useful, and how to keep its tools, permissions and outcomes under control.
2026-10-15 · 2 min read
Model Context Protocol (MCP): what it is and when to use it
A developer-focused guide to MCP servers, tools, resources and prompts, with practical advice on permissions and secure integrations.
2026-10-15 · 2 min read
A2A protocol and MCP: how they work together
Understand the difference between the Agent2Agent (A2A) protocol and MCP, and choose the right integration boundary for AI agents.
2026-10-15 · 2 min read
Agent Communication Protocol (ACP): what changed and how it relates to A2A
A clear guide to the Agent Communication Protocol (ACP), its A2A convergence, and why ACP can also mean a commerce protocol.
2026-10-15 · 2 min read
Agentic Commerce Protocol (ACP): how AI-agent checkout works
A practical introduction to the Agentic Commerce Protocol, agent-assisted checkout, merchant control and payment boundaries.
2026-10-15 · 2 min read
The agentic economy: what has to work before agents can trade
A grounded look at the agentic economy: task markets, identity, reputation, verification, payments and human accountability.
Work with me
Tell me what you are building
Share the goal, what you need help with and any timing or budget context. I will reply to discuss scope and next steps.
Prefer email? hello@muhaimin.dev