Available for AI, ML, and full-stack roles

Abdullah Al Muhimine(Tousif)

AI Full-Stack Engineer

I build practical intelligent systems: LLM agents, retrieval pipelines, ML models, secure APIs, and product-grade web applications. My strongest lane is turning AI ideas into usable software with clean interfaces and reliable backend logic.

View projectsDownload resumes
6+

AI and full-stack projects

15%

LLM reasoning accuracy gain

90%

Flood model accuracy

3.44

CSE CGPA

Abdullah Al Muhimine

tousif@portfolio:~$

shipping AI reasoning systems, retrieval apps, and full-stack products from Dhaka.

Four ways to read my profile

Same engineer, different role focus.

AF Resume

AI Full-Stack Allrounder

Best for roles that need one engineer who can connect LLM reasoning, production web apps, APIs, databases, and deployment.

Systems, AI products, backend, frontend, deployment
Resume A

AI Engineer

Focused on LangChain, LangGraph, Ollama, fine-tuning, PGVector retrieval, and AI application workflows.

LLMs, RAG, multi-agent systems, model evaluation
Resume F

Full-Stack / Backend Engineer

Built for backend and full-stack roles involving React, Next.js, Supabase, PostgreSQL, JWT auth, and API security.

Node.js, FastAPI, REST APIs, auth, databases
Resume M

ML / DL Engineer

Highlights supervised learning, feature engineering, CNNs, sequence modeling, multimodal learning, and research methods.

ML, deep learning, research, model validation

Selected work

AI systems with real application layers.

PorteHobe AI screenshotCapstone AI Reasoning

15% reasoning accuracy gain

PorteHobe AI

A math reasoning AI capstone built around open-source LLM refinement, benchmark-driven evaluation, and task-specialized reasoning workflows.

  • Fine-tuned Gemma 3:4B with supervised instruction tuning for mathematical reasoning.
  • Benchmarked against Mathstral and Gemini-style baselines with GSM8K evaluation.
  • Improved reasoning accuracy by 15% through dataset curation and model refinement.

Gemma 3:4B, Ollama, LangGraph, GSM8K, Python, LLM fine-tuning

Code
Data Structure Learning LLM screenshotAI Education Platform

LLM tutor with retrieval memory

Data Structure Learning LLM

An AI-powered DSA tutor that explains code, visualizes algorithms, and serves context-aware educational content from a backend-first architecture.

  • Built LLM reasoning workflows with LangGraph and MCP Server.
  • Designed APIs for context-aware query handling and vector-based retrieval.
  • Created interactive learning flows for algorithms and data structures.

Python, FastAPI, LangGraph, MCP Server, Supabase, PGVector, TypeScript

LiveCode
Bouncy Book Store screenshotFull-Stack E-commerce

Secure commerce workflow

Bouncy Book Store

A full-stack bookstore with role-aware access, authentication, cart flows, inventory-aware data, and a polished responsive storefront.

  • Architected the application with React, Node.js, and Supabase PostgreSQL.
  • Designed REST APIs with JWT authentication and role-aware access control.
  • Connected product, cart, and inventory workflows across frontend and backend.

React, Next.js, Node.js, TypeScript, Supabase, PostgreSQL, JWT

LiveCode
Flood Forecasting AI System screenshotML + Conversational AI

90% flood prediction accuracy

Flood Forecasting AI System

A hybrid flood forecasting system that combines supervised ML prediction with conversational LLM explanations for user-facing insight.

  • Developed a predictive model with time-series forecasting and 90% accuracy.
  • Engineered features from localized environmental datasets for prediction stability.
  • Integrated forecast outputs with an explanation engine and visual interface.

Python, Streamlit, LangChain, Scikit-learn, Gemma2:2B, JSON datasets

Code
Physics RAG Chatbot screenshotRetrieval-Augmented Assistant

Retrieval-first reasoning

Physics RAG Chatbot

A physics assistant that uses semantic retrieval before answering, reducing hallucinations and grounding explanations in relevant context.

  • Developed a RAG system using vector search for context-grounded reasoning.
  • Reduced hallucinations through retrieval-first answering logic.
  • Built structured workflows for physics problem explanation and follow-up answers.

Python, Streamlit, LangChain, PGVector, Hugging Face embeddings

Code
Multimodal Sentiment Analysis screenshotDeep Learning Research

Image + video emotion modeling

Multimodal Sentiment Analysis

A deep learning research project for detecting emotional sentiment from image and video inputs with refined temporal preprocessing.

  • Architected CNN-based and sequence models for multimodal sentiment detection.
  • Engineered temporal data preprocessing pipelines for video-based features.
  • Evaluated model behavior across image and sequence representations.

Python, PyTorch, TensorFlow, CNNs, sequence modeling, video preprocessing

Code

Resume-backed toolkit

Skills across AI, ML, backend, and product UI.

LLMs & GenAI

RAGMulti-agent systemsAgent orchestrationLLM integrationReasoning evaluationGSM8K

ML & Deep Learning

Supervised learningFeature engineeringCNNsSequence modelingMultimodal learningScikit-learn

Frameworks

LangChainLangGraphOllamaTensorFlowPyTorchStreamlit

Backend

Node.jsFastAPIDjangoREST APIsJWT authenticationAPI security

Frontend

ReactNext.jsTypeScriptJavaScriptTailwind CSSHTML/CSS

Data & Tools

SupabasePGVectorPostgreSQLMySQLMongoDBGit/GitHub

Foundation

Experience and education from the resumes.

Dec 2025-April 2026 | Dhaka

Full-Stack Developer - FabricoMartBD

  • Built the admin dashboard for internal business operations.
  • Developed and deployed CMS web platforms and supporting infrastructure for an international brand launch.
2021-2025 | CGPA 3.44 | AI and Machine Learning track

North South University

B.Sc. in Computer Science and Engineering

2018-2020 | GPA 5.00

National Ideal College

Higher Secondary Certificate (HSC)

2016-2018 | GPA 5.00

Mohanogor Ideal School

Secondary School Certificate (SSC)

Resume library

Download the version that matches the role.

The portfolio presents you as one complete builder, while the downloads let recruiters choose the AI, full-stack, or ML/DL angle immediately.

AI Full-Stack AllrounderAF ResumeAI EngineerResume AFull-Stack / Backend EngineerResume FML / DL EngineerResume M

Let's connect

Have an AI product, model, or backend that needs to become real?

abdullahtsn13@gmail.comGitHubLinkedInDhaka, Bangladesh