Amanat Ullah Mahmud

GenAI Engineer | ML Developer

Resume

Hello, I am

Amanat Ullah Mahmud.

0

Excel
Project

2

PowerBI
Projects

1

SQL
Project

1

Python
Project

2

Machine Learning
Projects

1

Deep Learning
Project

3

GenAI
Projects

About Me

I'm a GenAI Engineer and ML Developer passionate about building intelligent solutions that solve real-world problems. With expertise in machine learning, deep learning, and cutting-edge GenAI technologies, I've developed 8+ production-ready projects spanning healthcare AI, e-commerce automation, and predictive analytics.
My journey combines technical depth with practical application - from building RAG systems for medical professionals to creating conversational AI chatbots with voice capabilities. I specialize in end-to-end ML pipelines using Python, scikit-learn, PyTorch, and deployment with Streamlit, while leveraging LangChain and HuggingFace for advanced GenAI solutions.

Key Skills

MySQL

Python

LangChain

LangGraph

scikit-learn

pytorch

mlflow

My Projects

Talent Recruiter Agent

Domain/Function: Human Resource

E-commerce AI Agent

Domain/Function: Ecommerce

MediAssist-AI-Agent

Domain/Function: Health Care

Fruit Freshness Detection

Domain/Function: Agriculture (and Food Technology)

Employee Segmentation and Profiling

Domain/Function: Human Resource

Beverage Price Range Prediction

Domain/Function: Beverage Industry

Nova Mart Promotional Campaign Analysis

Domain/Function: Campaign Analysis

E-commerce Website Performance Analysis

Domain/Function: Ecommerce

CRM and Sales Analysis

Domain/Function: Customer Relationship Management and Sales

Ecommerce Sales Analysis

Domain/Function: E-commerce Sales and Promotions

Social Media Analysis

Domain/Function: Social Media Management

My Experience

Awards & Certificate

Data Analytics Bootcamp 5.0: Job Placement Support + AI Automation & Data Engineering Basics

Gen AI & Data Science Bootcamp: With Practical Job Placement Support & Virtual Internship

Let's Connect

Feel free to get in touch with me. I am always open to discussing new projects, creative ideas or opportunities to be part of your visions.

https://www.linkedin.com/in/amanat-mahmud

GitHub

https://github.com/amanat-mahmud

GenAI Project

Talent Recruiter Agent

Domain/Function: Human Resource

A smart AI recruitment assistant that automates resume screening and technical interviews. Candidates upload resumes that are instantly analyzed against job descriptions. Qualified applicants proceed to an AI-conducted interview with dynamic question-answering capabilities.

View Project
on Streamlit

Other Tools

LangGraph

The main interface of the app.

Graph Architecture of the agent.

Project Details

Imagine a scenario:
Abdullah, a software engineer, applies for a Machine Learning role at a fast-growing startup. Instead of waiting weeks for human screening, he:

1. Uploads his resume PDF
2. Receives instant feedback about Job Description match
3. Completes a live technical interview with AI

Within minutes, the recruiter gets a full candidate report including skills alignment analysis and interview responses - accelerating hiring while reducing bias.

Project video

GenAI Project

E-commerce AI Agent

Domain/Function: Ecommerce

A smart conversational AI agent that revolutionizes online shopping through natural language interactions. Users can ask questions about products or general inquiries using both text and voice input, receiving instant, contextually relevant responses powered by advanced semantic search and routing.

View Project
on Streamlit

Other Tools

LangChain

Agent answering both faq questions and voice prompt questions.

Agent fetching related products to user query.

Project Details

Imagine a scenario:
Ayesha, a busy mom, needs new running shoes but doesn't have time to browse through hundreds of products. Instead of spending hours scrolling through pages, she simply asks:

"Show me women shoes under 2500 taka with good ratings"

Within seconds, the AI agent understands her intent, searches the product database, and presents her with the perfect matches along with ratings, prices, and direct purchase links - making her shopping experience effortless and efficient.

Project video

GenAI Project

MediAssist-AI-Agent

Domain/Function: Health Care

View Project
on Streamlit

Other Tools

LangChain

The agent doesn't give any answer when asked non health related questions.

Answers healthcare related questions with pubmed source.

Project Details

A user-friendly agent(LangChain) designed for effortless information retrieval from pubmed article. Users can input search term and ask questions to receive relevant insights from the healthcare domain. (But it's features can be extended to any domain.)

Imagine a scenario:

Dr. X, a busy doctor at Metro Health, has a Type 2 diabetic patient asking about intermittent fasting but lacks time to review dozens of conflicting studies
Instead of spending hours researching, Dr. X can now query MediAssist AI "Is 16:8 intermittent fasting safe for Type 2 diabetics?" and receive a concise, peer-reviewed summary in seconds, enabling informed clinical decisions

Project video

Deep Learning Project

Fruit Freshness Detection

Domain/Function: Agriculture (and Food Technology)

View Project
on Streamlit

Other Tools

Pytorch

Detecting the fruit quality.

Detecting the fruit quality

Project Details

A fruit supplier company need to supply fresh food to supermarkets and local farmers' markets, striving to deliver high-quality goods to their customers. They are facing several operational issues due to inconsistent manual quality inspections.So they plans to leverage AI to address this problem by automating the freshness inspection process.
They supply 8 kinds of fruits/vegetables: - Banana - Lemon
- Lulo
- Mango
- Orange
- Strawberry
- Tamarillo
- Tomato

Machine Learning Project

Employee Segmentation and Profiling

Domain/Function: Human Resource

In this project applied K-means clustering to an HR dataset to uncover four distinct employee personas based on tenure, role, income, and satisfaction metrics. By leveraging targeted feature engineering, PCA-based 2D visualization, cluster evaluation (silhouette, Calinski–Harabasz, Davies–Bouldin),

Recommendation and clusters.

Project Details

Cluster 0 – Early-Career R&D Technicians

Clear tech career path & flexible hours
Adjust entry-level salaries
Cluster 1 – Entry-Level Sales Reps

Strengthen onboarding & travel perks
Offer fast-track promotion & mental health support
Cluster 2 – Senior R&D Managers

Enforce vacation & hire support staff
Launch executive wellness programs
Cluster 3 – Mid-Career Sales Executives

Fast-tracked pay raises & retention bonuses
Accelerate promotion to management

Machine Learning Project

Beverage Price Range Prediction

Domain/Function: Beverage Industry

This project develops a Machine Learning-based Price Prediction System that analyzes customer survey data to predict the most suitable price range for individual customer segments.

View Project
on Streamlit

Other Tools

Scikit-Learn

mlflow

optuna

Using this ui we can put peoples preference and get the price range.

WE get the price range with the confidence of the model.

Project Details

Codex, a leading beverage company, is planning to launch an innovative new product in the competitive beverage market. To ensure successful market penetration and optimal pricing strategy, the company conducted comprehensive customer surveys to understand consumer preferences, purchasing behavior, and price sensitivity. This project develops a Machine Learning-based Price Prediction System that analyzes customer survey data to predict the most suitable price range for individual customer segments.

Project video

Python Project

Nova Mart Promotional Campaign Analysis

Domain/Function: Campaign Analysis

Analyzed Diwali 2023 and Sankranti 2024 campaign datasets with Python and Pandas to quantify sales uplift, IR%, ISU%, and store-/category-level performance. Derived key insights and formulated targeted recommendations to optimize Nova Mart’s future promotional strategies.

View Project
Engagement

View Project
on GitHUB

Thumbnail

Project Details

E-commerce Website Performance Analysis

Domain/Function: Ecommerce

This project presents a comprehensive statistical analysis of an e-commerce platform’s visitor session data to uncover key drivers of conversion performance and identify opportunities for optimization. By leveraging rigorous hypothesis testing, confidence interval estimation, and visualizations.

View Project
on GitHUB

Business Questions.

Key findings.

Project Details

PowerBI Project

CRM and Sales Analysis

Domain/Function: Customer Relationship Management and Sales

Data-driven project optimizing sales performance and revenue forecasting through comprehensive analysis of lead distribution, conversion rates, and sales pipeline health.

View Project
Engagement

View Project
on GitHUB

Overview page

Details Page

Project Details

❓Problem Statement
A company aims to evaluate its CRM data and sales pipeline for leads registered over the last five months. The task is to build a comprehensive analytic report that provides insights into lead distribution across countries, industries, and organization sizes. The report assesses the health of the sales pipeline, forecasts potential income over the next few months, and compares sales agent performance.

🎯 Objective
The primary objective is to provide actionable insights that will optimize sales performance, improve lead conversion rates, and enhance revenue forecasting accuracy. Specific goals include:
Identify top-performing sales agents and their strategies
Evaluate lead conversion rates across different dimensions
Assess sales pipeline health month-over-month
Analyze impact of response time on conversion rates
Examine trends in average deal values
Develop a robust revenue forecasting model
Investigate relationships between organization size, sales cycle.

Project video

Live Interactive Dashboard

PowerBI Project

Ecommerce Sales Analysis

Domain/Function: E-commerce Sales and Promotions

Comprehensive e-commerce sales and promotions analysis (2019) using Power BI. Examines user behavior, product performance, and promotional impact across brands, categories, and states to optimize sales strategies and profitability.

View Project
Engagement

View Project
on GitHUB

Overview page

Day Time Analysis Page

Project Details

❓ Problem Statement
In a competitive market, understanding the factors driving sales performance, profitability, and customer satisfaction is crucial for sustainable growth. Our organization faces challenges in identifying top performing products, optimizing inventory, optimizing sales and promotions strategies necessitating a deeper understanding of user behavior. Additionally, we need to understand the financial impact of discounts to enhance our profit margins.

🎯 Objective
Analyze sales and promotions data to identify key trends and insights, segmenting analysis by Overview, Product, Day/Time, Brand, and State for targeted insights. Aims include identifying best-selling brands, assessing financial health, understanding various factors' impact on sales, and providing actionable recommendations.

🔎 Key Findings
Total sales: $5.37M
Purchase ratio: 9.11%
Best-selling brand: Apple
Average rating: 2.51 (significantly low)
78% of customers only visit without purchasing

Project video

Live Interactive Dashboard

SQL Project

Social Media Analysis

Domain/Function: Social Media Management

Data-driven project focusing on optimizing content strategy and user engagement through hashtag analysis, user influence metrics, and regional content preferences.

View Project
Engagement

View Project
on GitHUB

Data model

Data Model

Project Details

❓ Problem Statement
While the social media platform shows active user engagement and clear trends in content preferences, there are opportunities to optimize hashtag usage, leverage influential users, and enhance regional content strategies. The platform faces challenges in balancing the influence of top users, capitalizing on trending topics, and ensuring consistent engagement across different content types and geographical regions. Additionally, there's a need to understand and replicate the success factors of highly bookmarked posts and active commenters to boost overall user interaction and content visibility.