Available for Software & AI Engineering Roles

Hello, I'm👋

Mohammad
Ali Nabil.

Software Developer & AI Systems Engineer

Computer Science undergraduate at North South University and former GenAI intern at The Data Island. I build production-grade software, autonomous multi-agent pipelines, RAG systems, and full-stack web applications with Python, FastAPI, and modern frontend tools.

Education
BSc in CSE
North South Univ.
Experience
The Data Island
GenAI Team Intern
Location
Dhaka, Bangladesh
Open to Remote
nabil_profile.ts
TypeScript 5.0
const engineer = {
  name: "Mohammad Ali Nabil",
  role: "Software Developer & AI Builder",
  location: "Dhaka, Bangladesh",
  passion: "Building hardcore systems, not just calling APIs",
  focus: [
    "Autonomous Multi-Agent Systems (LangGraph)",
    "LLM & Large Multimodal (LMM) Fine-Tuning & Model Training",
    "Retrieval-Augmented Generation (Dense/Sparse RAG)",
    "Scalable Full-Stack Web Backends (FastAPI, React)",
  ],
};
Building practical systems & modern apps@THENABILMAN
INTERACTIVE SHELL & RAG //

Execute Terminal.

Query the vector knowledge base directly or run live agent simulations through an interactive developer shell.

VECTOR RETRIEVALPinecone DB
INFERENCE MODELNex-N2.5 Mini
EMBEDDINGS1024-dim llama
STATUSOnline (< 350ms)
Commands:
Sample Queries:
guest@runtime: ~/portfolio-rag-agent

Interactive RAG Shell [Connected to Vector Store & Gemma LLM]

Type any technical question, run commands, or select presets above.

$guest@runtime:~#system --info
  _  _  _   ___ ___ _  
 | \|  |/_\  | _ ) _ _| | 
 | .` / _ \ | _ \| || |__
 |_|\_/_/ \_\___/___|____|

SYSTEM ARCHITECTURE

Autonomous RAG & Multi-Agent Runtime

OS / KERNEL:Agentic Linux x86_64 (FastAPI + Next.js)

VECTOR STORE:Pinecone Serverless (1024-dim llama-text-embed-v2)

INFERENCE LLM:Google Gemma-4-31B-IT via OpenRouter

AGENT WORKFLOW:LangGraph StateGraphs + MCP Tool Schema

SYSTEM STATUS:🟢 Online & Responding (< 350ms)

$
CAREER & ACADEMIC TIMELINE //

Experience & Education.

Applied engineering experience in GenAI workflows combined with rigorous foundational computer science training.

PRACTICAL TRACK
GenAI Engineering
Production pipelines & agents
ACADEMIC DEGREE
BSc in CSE
North South University
TECHNICAL FOUNDATION
Theory + Systems
Algorithms, AI & Backends
Industry ExperienceCompleted
Nov 2025 – Jan 2026Remote / Dhaka, Bangladesh

AI Engineer Intern

The Data IslandGenAI Team

Engineered end-to-end AI automation workflows, high-precision structured data extraction pipelines, and agentic reasoning verification frameworks.

Key Impact & Contributions:
Architected scalable AI automation workflows and LLM API integrations for high-precision structured data extraction.
Engineered robust Python web scraping and validation pipelines with automated data cleaning and normalization routines.
Designed asynchronous batch execution architectures to process heavy data workflows with exponential retry logic.
Debugged and optimized complex agent reasoning loops, reducing hallucinations and improving state machine reliability.
Core Competencies Applied:
#Python#LLM APIs#LangGraph / LangChain#Structured Extraction#Web Scraping#Batch Pipelines#Agent Debugging#Data Normalization
Academic FoundationActive Track
Undergraduate ProgramDhaka, Bangladesh

BSc in Computer Science & Engineering

North South University (NSU)Department of Electrical & Computer Engineering

Comprehensive computer science curriculum balancing rigorous algorithmic theory with modern applied machine intelligence and backend software engineering.

Key Impact & Contributions:
Core focus on Artificial Intelligence, Machine Learning, Data Structures & Algorithms, and Distributed Systems.
Hands-on research and course projects in Computer Vision, Natural Language Processing, and scalable Python/C++ backends.
Applied theoretical concepts into production-grade systems, agentic architectures, and vector search RAG pipelines.
Targeting future academic trajectory toward advanced research in machine intelligence, deep architectures, and quantum computing.
Core Competencies Applied:
#Data Structures & Algorithms#Artificial Intelligence#Machine Learning#Distributed Systems#Database Systems#Operating Systems#C++#Object-Oriented Architecture
FEATURED_PORTFOLIO //

Featured Systems & Applications.

Production-grade agentic architectures, low-latency streaming pipelines, and full-stack software deployments.

★ FLAGSHIP SYSTEMFull-Stack & Multi-AgentActive Production

FITMAN

Full-Stack AI Fitness & Nutrition Coaching Platform

Production-deployed fitness and nutrition platform orchestrating LangGraph multi-agent state machines, Groq streaming inference, and real-time FastAPI backends for personalized daily coaching and health telemetry adherence.

CORE ARCHITECTURE DELIVERABLES
Engineered LangGraph agent state machines with multi-turn memory for adaptive workout & macro planning.
Integrated Groq LLaMA-3 for low-latency streaming responses (< 300ms Time-to-First-Token).
Designed Supabase PostgreSQL schema with strict Row-Level Security (RLS) policies and health telemetry tracking.
Developed a glassmorphic mobile-first React frontend with real-time interactive charting and habit tracking.
END-TO-END EXECUTION PIPELINE
1Client UI (React/Vite)
2FastAPI Gateway
3LangGraph Multi-Agent DAG
4Groq LLaMA-3 (Sub-300ms)
5Supabase PostgreSQL
TECHNOLOGY STACK
#LangGraph#FastAPI#Python#Groq LLaMA-3#Supabase#PostgreSQL#React#TypeScript#Tailwind CSS
ENGINEERING REPOSITORIES & AGENT SYSTEMS (6)
Agentic & RAGCompleted

End-to-End Autonomous Agent

Self-Reflective Multi-Step Reasoning Engine

Autonomous agent framework equipped with persistent Pinecone vector memory, dynamic tool calling capabilities, web retrieval, and self-correcting evaluation loops.

Built multi-step planning loops that decompose ambiguous user queries into verifiable sub-tasks.
#Python#LangChain#Pinecone#Groq / OpenAI API#Pydantic
Agentic & NLPCompleted

AI Resume Analyzer

LLM-Powered Semantic Parsing & Job Fit Scoring

Intelligent resume parsing and ATS optimization engine using LLMs, vector embeddings, and semantic similarity scoring to evaluate candidate skills, match job descriptions, and deliver actionable CV insights.

Engineered structured JSON entity extraction from complex multi-page PDF/DOCX resumes using constrained LLM schemas.
#Python#LLMs#FastAPI#Streamlit#LangChain
Voice & Real-TimeCompleted

Real-Time AI Voice Agent

Low-Latency Conversational Voice Pipeline

Full-duplex conversational voice agent integrating real-time streaming Speech-to-Text (STT), low-latency LLM reasoning, and natural neural Text-to-Speech (TTS) for human-like auditory interaction.

Implemented streaming audio pipeline with Voice Activity Detection (VAD) for fluid turn-taking.
#Python#Whisper STT#Neural TTS#WebSockets#FastAPI
Voice & Bengali NLPCompleted

Bangla AI Voice Agent

Bengali Speech Recognition & Speech Synthesis

Native Bengali conversational AI agent supporting end-to-end Bangla Speech-to-Text, contextual LLM inference in Bengali, and low-latency Bangla neural speech generation for localized verbal assistance.

Integrated specialized Bengali ASR/STT models for accurate native speech recognition.
#Bengali NLP#Bangla STT#Bangla TTS#Python#PyTorch
Voice & Real-TimeActive Prototype

LiveKit MCP Voice Assistant

Bidirectional WebRTC Voice Assistant with MCP

Voice-first AI assistant utilizing LiveKit WebRTC for full-duplex speech communication, Model Context Protocol (MCP) server tools, and low-latency neural speech synthesis.

Achieved sub-200ms voice roundtrip latency using LiveKit WebRTC audio transports.
#LiveKit WebRTC#Model Context Protocol#Whisper STT#Cartesia TTS#Python
Computer VisionCompleted

AI Vision & OCR Automation

Real-Time Video Stream & Document Extraction Engine

Computer vision automation pipeline leveraging OpenCV video filtering, Hugging Face vision transformers, and OCR models for real-time document extraction and object classification.

Processed live video streams through OpenCV filtering pipelines for rapid artifact boundary detection.
#OpenCV#Hugging Face#PyTorch#EasyOCR#Python
KNOWLEDGE_ARCHITECTURE // SYSTEM SKILLS

Technical Knowledge Graph.

Complete engineering hierarchy: from foundational Python runtime, asynchronous backends, and databases to tensor calculus, deep neural transformers, autonomous state graphs, and LLM/LMM fine-tuning.

CORE DOMAINS7 Engineering Layers
TECHNICAL SKILLS60 Verified Competencies
SPECTRUMTheory + Production RAG
EXECUTION STATUS Production Tested
system@runtime:~$ tree --skills --hierarchy
GRAPH_ONLINE
Python(10 skills)

Deep mastery of the Python runtime, metaprogramming, OOP design patterns, and package tooling.

CORE RUNTIME & OOP
├──syntax
[Core]
├──functions
[Core]
├──classes / objects
[Core]
├──modules / imports
[Core]
├──exceptions
[Core]
├──files
[Core]
├──iterators / generators
[Advanced]
├──decorators
[Advanced]
├──typing
[Advanced]
└──package management
[Production]
Backend(8 skills)

Asynchronous microservices, high-concurrency API design, and network protocol routing.

ASYNC SERVICES & APIS
├──HTTP
[Core]
├──REST
[Core]
├──JSON
[Core]
├──APIs
[Production]
├──FastAPI
Used in FITMAN
├──async / await
[Advanced]
├──authentication
[Production]
└──middleware
[Advanced]
Databases(7 skills)

ACID transactions, relational schemas, indexing strategies, and Supabase / PostgreSQL optimization.

PERSISTENCE & SCHEMAS
├──SQL
[Core]
├──PostgreSQL
Supabase RLS
├──relationships
[Core]
├──joins
[Core]
├──indexes
[Advanced]
├──transactions
[Advanced]
└──ORM
[Production]
Software Engineering(9 skills)

Systems engineering best practices, Linux environments, testing suites, and clean architecture.

RIGOR & PRODUCTION CRAFT
├──Git
[Production]
├──Linux
[Production]
├──project architecture
[Advanced]
├──testing
[Production]
├──debugging
Agent Debugging
├──logging
[Production]
├──environment variables
[Core]
├──dependency management
[Core]
└──clean code
[Production]
AI Fundamentals(6 skills)

Mathematical foundations of machine learning, tensor algebra, probability, and classical modeling.

MATH & MACHINE LEARNING
├──NumPy
[Production]
├──Linear Algebra
[Core]
├──Probability
[Core]
├──Calculus
[Core]
├──Machine Learning
[Advanced]
└──Neural Networks
[Advanced]
Deep Learning & Model Training(9 skills)

PyTorch tensor operations, backprop calculus, sequence models, multi-head attention, and LLM model training workflows.

NEURAL ARCHITECTURES & TRAINING
├──Backpropagation
[Core]
├──PyTorch
[Production]
├──Transformers Architecture
[Production]
├──LLM Model Training
Pretraining / SFT / DPO
├──CNN & Vision Encoders
[Advanced]
├──RNN/LSTM
[Advanced]
├──Attention Mechanisms
[Advanced]
├──Model Quantization
AWQ / GGUF / bitsandbytes
└──Distributed Training
DeepSpeed / FSDP
AI Engineering & LLM/LMM Tuning(11 skills)

Autonomous state graph agents, dense/sparse vector retrieval, LLM & LMM (multimodal) fine-tuning (LoRA/QLoRA), and model alignment.

AGENTIC SYSTEMS, RAG & FINE-TUNING
├──LLM & LMM Fine-Tuning
LoRA / QLoRA / PEFT
├──Unsloth & Hugging Face TRL
Sub-Hour LoRA
├──Dataset Curation & SFT
Instruction Tuning
├──RAG Architecture
Pinecone 1024-dim
├──embeddings
[Production]
├──tool calling
MCP Schema
├──agents
[Production]
├──LangGraph
FITMAN DAG
├──evaluation & benchmarking
[Production]
├──deployment & serving
vLLM / Ollama
└──multimodal / vision
LiveKit & OpenCV