Moshiur R.
AI Full Stack Developer | Next.js, Python, Multi agent, Enterprise RAG
Hi! I'm Moshiur Rahman Sifat, a full stack web developer with 4+ years of experience building and shipping complete web products. I build fast React and Next.js front ends, Node.js, NestJS and Python (FastAPI) back ends, and add AI where it earns its place: RAG chatbots over your own documents, LLM agents and multi-agent assistants. I'm also comfortable training and evaluating machine learning and deep learning models when a product needs one. WHAT I CAN BUILD FOR YOU • Web apps and SaaS: React, Next.js, TypeScript, JavaScript (ES6+), Tailwind CSS, shadcn UI, dashboards, admin panels, landing pages, responsive UI • Back ends and APIs: Node.js, Express, NestJS, Python FastAPI, REST APIs, authentication, third-party integrations • Databases: PostgreSQL, MongoDB, MySQL, Supabase, vector search with Qdrant • AI features: RAG chatbots with hybrid search, citations and role-based access; LLM agents with tool calling; multi-agent assistants (LangGraph); natural-language-to-SQL; prompt-injection guardrails; evaluation test sets so you can measure quality • Machine learning and deep learning: PyTorch and Transformer models for classification, ranking and recommendation, image similarity search (FAISS), ONNX export • Deployment: Vercel, Azure, Docker, Kubernetes RECENT PROJECTS (live demos and code are in my Portfolio section) 1. Enterprise RAG chatbot: answers only from documents each role may read; 0% leaks on held-out tests. FastAPI, Qdrant, React. 2. Multi-agent AI assistant: routes each question to one of five specialist agents; 96% routed correctly on a 59-question test set. LangGraph, Python. 3. Natural-language to SQL agent: ask a database questions in plain English and see the query and rows. 15/15 on a fixed demo eval. 4. Harmful content detection: DistilBERT moderation system with a human review queue, FastAPI, React, deployed on Azure. 5. Personalized news feed: multi-task neural ranker with retrieval, ranking and re-ranking; 1.8x the click-through rate of a chronological feed in a simulated A/B test. GOOD FIT IF YOU ARE • A startup that needs an MVP or a complete web app built end to end • A team adding an AI assistant, chatbot or document search to an existing product • A business that wants a clean React or Next.js front end with a solid API behind it • Looking to fix, speed up or upgrade an existing web app HOW I WORK Clear communication, clean and tested code, measurable results and honest limitations. I take projects from design to deployment and I'm available for both short-term fixes and long-term work. Let's build something useful together!
Earnings and projects
Last 6 months ÷ 6
Last 6 months: $4.9k
Billed in all: $11,195$11k
$2.5k in the last 6 months
Across 5 contracts
Earnings, last 30 days
Skill rankings: Top 11% in ExpressJS
Ranked by MRR among everyone on Upwork with the skill.
- #347Deep Learning2,743 freelancers · 🇧🇩 #7Top 13%🇧🇩#7 of 49
- #509Chatbot Development3,106 freelancers · 🇧🇩 #8Top 16%🇧🇩#8 of 80
- #528NestJS3,757 freelancers · 🇧🇩 #9Top 14%🇧🇩#9 of 126
- #618ExpressJS5,651 freelancers · 🇧🇩 #19Top 11%🇧🇩#19 of 258
- #619Large Language Model2,726 freelancers · 🇧🇩 #4Top 23%🇧🇩#4 of 37
- #755RESTful API6,064 freelancers · 🇧🇩 #8Top 12%🇧🇩#8 of 213
- #794LangChain3,821 freelancers · 🇧🇩 #6Top 21%🇧🇩#6 of 77
- #857Tailwind CSS6,978 freelancers · 🇧🇩 #29Top 12%🇧🇩#29 of 437