Project Highlights:

• Input Simplification Better UX: The process begins with an intuitive interface where users only input a YouTube URL. This simplicity required careful planning to ensure seamless user experience while managing complex backend processes.
• Efficient Transcript Retrieval: Implemented an AWS Lambda function to dynamically extract YouTube transcripts. This involved analyzing YouTube's transcript API, optimizing for latency, and ensuring scalability to handle multiple requests simultaneously.
• Prompt Engineering: Crafted and refined several specialized prompts for the GPT model, combining creativity with rigorous testing. These prompts were designed to extract nuanced insights from the transcript and generate SEO-optimized, high-quality articles tailored to different content styles.
• SEO Optimization: Integrated the Yoast SEO plugin to automatically assess the generated articles. Achieving green marks for 🟢SEO score, 🟢Readability, and 🟢Inclusive language required meticulous analysis of the Yoast algorithm and iterative adjustments to content structure and metadata.
• AI-Powered Image Generation: Designed an AI-based image generation pipeline to produce visually appealing featured images. The system uses different models depending on user preferences (basic, medium, or high quality), requiring a deep understanding of image processing and user needs.
• SEO-Driven External Link Integration: Developed a mechanism using Google Custom Search Engine (CSE) API to identify and incorporate authoritative external links. This step involved keyword mapping and automated relevance filtering to enhance SEO rankings while maintaining content authenticity.
• Engaging Users: Implemented a Socket.io-based real-time logging system to show users the progress of each step. This required efficient communication between client-side scripts and server-side processes.
• Dynamic Article Expansion: Engineered a process to expand AI-generated articles, ensuring they meet long-form content requirements without compromising quality. This involved recursive content enrichment using AI-driven models for tone consistency and depth.
• Streamlined Publishing Workflow: Integrated the WordPress REST API for direct uploading of articles and images. This required analytical thinking to map complex metadata structures, such as Yoast SEO configurations, into a seamless publishing flow.
• Cloud-Based Media Storage: Developed an image upload and management system using AWS S3 via the Media Cloud plugin. This step involved designing a secure, scalable architecture for storing and retrieving large volumes of media files efficiently.
• Editable Previews for User Control: Incorporated a markdown editor for article previews, giving users flexibility for final edits. Building this feature required balancing performance and rich-text editing capabilities.
• End-to-End Architecture Design: The software's backbone is a robust full-stack architecture utilizing technologies like Next.js, Python, Express.js, AWS Lambda, DynamoDB, and EC-2 deployment. Each decision was driven by analytical trade-offs between performance, scalability, and maintainability.

Business Impact:

• Automated Workflow: Can do in 1 minute that needs 7-8 hours for a human, enabling creators to focus on growth and strategy.
• SEO Mastery: Combines cutting-edge AI, metadata optimization, and external link integration to ensure top rankings.
• Scalability: Perfect for businesses aiming to publish large volumes of content with minimal effort.
• Cost-Efficient: Reduces dependency on expensive content creation services by automating the entire process.

Built with a deep understanding of full-stack development and scalable cloud solutions, this project highlights my ability to craft efficient, business-driven applications.

Key points :
• User can add or edit a word
• User can add/edit pronunciation of the word
• User can add/edit importance level of the word
• User can add/edit definition of the word
• User can add/edit parts of speech (one or more) of the word
• User can add/edit side note of the word
• User can add/edit mnemonic as well as mnemonic note about the word
• User can add/edit derived words as well as side note of the word
• User can search for existing synonym. add/edit synonym of the word
• User can search for existing antonym . Can add/edit antonym of the word
• User can search word to find sentences containing that word
• To prevent data loss, whole database is backed up twice a day using cron job and save to dropbox using dropbox API

Key points :

Key points :
• Weather data is taken from wUnderground API
• More weather data is scrapped from website, and then parsed
• Weather forecasting message is sent to registered farmers' mobile numbers using SMS API

Key points :
• User can search for existing or create community anew using the name of their institute/department. Any user can add member by sharing link like Google classroom
• Users can post, all members will be notified by email
• Members can create directories, upload files, share file links like Google Drive. A desktop application (written in c#) sync all files in user's computer (mirror files, folders)
• Teacher can join and be take role of teacher/admin. And s/he will have more privilege in controlling the community
• Files are stored in Amazon S3
• And android app also mirrors files, folders. And shows posts from website
• Course management system for teachers
• Question answer forum

Key points :
• Volunteer donor can register and add and edit their personal information
• Donor can be found searching by their blood group who donated blood more than 3 months ago

Key points :
• Basic e-commerce functionality
• Used Laravel, Vue.js. Used mailgun for email service

details

• I am currently working on it.
• Source code will be shared as open source project, in github, soon

How it works :
• When bKash payment confirmation SMS arrives to phone, using Java code that SMS is read immediately
• Then the SMS is parsed, which mobile number it comes from and arrival time and Transaction ID etc. are parsed.
• That information is immediately sent to server. The app turns ON internet if not already ON
• This is how payment can be verified within few seconds

Key points :
• Search both Bengali and English words
• Obligatory revision system : user can set after reading how many words revision should be prompted. By default if 20 words is read by user it will prompt user to revise those
• MCQ test : Quiz questionnaire are generated by algorithm. Quiz system is developed using Laravel (php) framework
• User is (push) notified read words now and then weather s/he recalls the meaning. App analyzes user behavior, and accordingly decides when and which word is to be shown
• Tons of raw data crawling as well as parsing were done to prepare the words information

Key points : This app is connected to the previous project (Lets Organ).

Key points :
• Daily Star
• Jugantar
• Ittefaq
• Kaler Kantho
Collects all the headlines from those newspapers. A HTML page is created as showing in the screenshot attached below. That page is just a list of headlines. So that user can skim through all the headlines and read only desired news. Reading this way s/he can read all important news of the day from 4 newspapers in significantly less amount of time.
Websites are crawled using jsoup library. This time saver software is developed for PC (having java installed).

• Using facebook graph API takes name, email, date of birth etc. from user profile
• Using some logic generate random results with user's image

• Admin can add, edit, delete news article
• News are shown in different categories
• Detailed article is opened in new tab on headline click
• Admin can create, edit, delete news category
• Most viewed news are shown at a tabbed panel
• Visitors can comment by registering/loging in

Key points :
• Purpose : To build a rich vocabulary learning app
• Search google using java to gather data of word's definition, synonyms
• Search cambridge website using java to collect data of word's definition, example sentences
• Search some other websites to gather some other data about thousands of words
• And save all the processed data in .txt and/or .json files

Key points :
• Takes points in this format (x1,y1), (x2,y2)...
• Saves any data once used to plot.
• Graph can be re-plotted using that data
• Graph can be saved as image
• Graph can be printed directly from right-click menu
• Zoom-in zoom-out option in X and Y both axis

Key points :
• Used Java, Swing GUI builder

Key points :
• With the help of bash programming the software runs at startup
• It saves anything is copied. So that anything copied once can be reused later
• Used PyQT for GUI building
• Sqlite for storing data

Key points :
• Connected to the project Lets Organ (previously mentioned)

Key points :

How it works :
• I developed an android app, which detects shake or takes command from ON/OFF button. It can detect volume change.
• It sends the command as json format to bluetooth module (HC-05).
• Bluetooth module sends it to Arduino via software serial.
• Arduino (UNO) then controls a servo motor (SG90). The servo is attached with a general fan regulator.
• The system automatically reduce fan speed after a used defined time at night as temperature falls at night
• Apart from shaking, also sticky ON/OFF button on notification bar for easy access
• Fan ON/OFF can be done by shaking the device

Key points :
• ESP8266 (NodeMCU v3) wifi module is used as controller
• DHT 11 temperature and humidity sensor is used to measure water temperature
• If target temperature is achieved, water heater is turned OFF
• If water temperature is measured less than target temperature, heater is turned ON
• Also a rice cooker is connected with this network. The rice cooker is turned ON/OFF as per scheduled time. I use it for my cooking :D

Key points :
• Everyday switched ON/OFF as per scheduled time