Mini 1: OCR Text Extractor with GCP Vision API & Streamlit
Overview:
Developed a Python tool with an interactive Streamlit web interface to extract text from images using the powerful Google Cloud Vision API.
Project Description
This project is a practical demonstration of leveraging cloud AI services for document processing. It provides a user-friendly web application where users can upload an image (JPG or PNG) and instantly receive the extracted text. The application also offers a feature to download the extracted text as a structured CSV file, bridging the gap between raw images and usable data.
Problem Statement
The process of manually transcribing text from images or scanned documents is tedious, time-consuming, and prone to error. While powerful Optical Character Recognition (OCR) technology exists, it often requires complex setup or integration. This project aimed to solve that problem by creating a simple, accessible, and highly accurate OCR tool that abstracts the complexity of a cloud-based AI service into a straightforward web interface, enabling anyone to digitize documents with just a few clicks.

Methodologies & Project Workflow
The project was executed in five distinct phases to ensure a structured and systematic development process:
Phase 1: Local Environment & Foundational Setup:
The initial phase focused on creating a standardized and isolated local development environment. This involved setting up a Python virtual environment for dependency management, initializing a Git repository for version control, and installing the core
google-cloud-visionlibrary.Phase 2: Google Cloud Platform Integration:
This phase involved configuring the necessary GCP services. The Cloud Vision API was enabled, and the local application was securely authenticated using Application Default Credentials (ADC) via the
gcloudCLI, eliminating the need to manage service account keys in the code.Phase 3: Core OCR Script Implementation:
A core Python script (
ocr_tool.py) was developed to handle the text extraction logic. This script used thegoogle-cloud-visionclient library to send image content to the API and process the returned text annotation.Phase 4: Interactive Web Interface Development:
To enhance usability, an interactive web interface was built using Streamlit. This involved creating a file uploader for images, displaying the results, managing application state, and implementing a feature to download the extracted text as a CSV file.
Phase 5: Final Documentation & Publishing:
The final phase involved creating a comprehensive
README.mdfile and publishing the complete project to a public GitHub repository. This ensured the project was well-documented and accessible as a portfolio piece.
Key Results
Project-Specific Outcomes: Developed a fully functional Streamlit web app for real-time OCR on uploaded images (PNG/JPG/JPEG) using Google Cloud Vision API's document_text_detection, enabling seamless text extraction, display, and download as .txt or .csv files—demonstrating end-to-end integration in a rapid 1-day prototype.
Efficiency and Usability: Achieved responsive performance via @st.cache_resource for API client caching, with live processing (spinners/alerts) completing in under a minute per image on local runs at http://localhost:8501.
Key Achievements: Resolved authentication hurdles (Google Cloud SDK/ADC) and corporate proxy issues for local development, showcasing practical skills in GCP integration and secure API handling without custom model training.
Model Insights: Leveraged the pre-trained Vision API for high-fidelity document text detection; qualitative tests on samples (e.g., receipt1.jpg) produced accurate extractions, though no quantitative metrics (e.g., accuracy) were evaluated—aligning with the project's focus on prototyping over benchmarks.
Key Features & Technical Details
Cloud-Powered OCR: Leverages Google Cloud Vision API for highly accurate and reliable text extraction from various image types.
Dual Interface: Offers both a command-line script for automated processing and an interactive web app built with Streamlit for ease of use.
Interactive Web UI: Allows users to easily upload JPG/PNG images, view the uploaded image, and see the extracted text in a clean text area.
Data Export: Includes a one-click download button to save the extracted text lines as a structured CSV file, ready for analysis.
Secure Authentication: Utilizes Application Default Credentials (ADC) for a secure, keyless authentication method to GCP services during local development.
Technical Stack
Cloud Platform: Google Cloud Platform (GCP)
AI Service: Cloud Vision API
Programming Language: Python
Web Framework: Streamlit
Core Libraries:
google-cloud-vision,pandasVersion Control: Git, GitHub
Challenges & Solutions
Challenge: Ensuring secure and seamless authentication between a local application and GCP APIs without hardcoding credentials.
Solution: Implemented Application Default Credentials (ADC) by configuring the
gcloudSDK (gcloud auth application-default login). This provided a secure and standard authentication mechanism that the Google Client Library could automatically use.
Challenge: The core OCR function was designed to work with file paths, but Streamlit's file uploader provides an in-memory byte stream, creating an incompatibility.
Solution: Integrated Python's
tempfilemodule to temporarily store the uploaded image data to a file. This generated a temporary file path that was passed to the existing OCR function, effectively bridging the web interface with the backend logic without significant refactoring.




Future Enhancements & Improvement
PDF & Multi-Page Support: Extend functionality to process multi-page PDF documents, extracting text from each page.
Cloud Deployment: Deploy the Streamlit application to a managed service like GCP's Cloud Run or Streamlit Community Cloud for public access.
Layout and Table Recognition: Enhance the processing to not only extract text but also recognize and preserve document structure, such as tables and paragraphs.
More Works
(GQ® — 02)
©2024
More Works
(GQ® — 02)
©2024

Project 1: Real-Time Fraud Detection API
2025 - Completed
API Development
XGBoost
FastAPI
AWS Cloud
Serverless Deployment
SHAP

Project 1: Real-Time Fraud Detection API
2025 - Completed
API Development
XGBoost
FastAPI
AWS Cloud
Serverless Deployment
SHAP

Project 2: Automated MLOps CI/CD Pipeline
2025 - Completed
MLOps
CI/CD
Docker
Spark
Serverless Deployment
AWS Cloud

Project 2: Automated MLOps CI/CD Pipeline
2025 - Completed
MLOps
CI/CD
Docker
Spark
Serverless Deployment
AWS Cloud

Project 3: QA Bot with Multi-turn RAG for e-Commerce
2025 - Completed
RAG
Multilingual
FastAPI
Gemini API
Hugging Face
Pinecone

Project 3: QA Bot with Multi-turn RAG for e-Commerce
2025 - Completed
RAG
Multilingual
FastAPI
Gemini API
Hugging Face
Pinecone

Mini 2: Serverless ETL Pipeline on GCP for CSV Automated Processing to BigQuery
2025 - Completed
Data Engineering
ETL
Serverless
GCP
BigQuery
Cloud Functions

Mini 2: Serverless ETL Pipeline on GCP for CSV Automated Processing to BigQuery
2025 - Completed
Data Engineering
ETL
Serverless
GCP
BigQuery
Cloud Functions

Mini 3: Sentiment Analysis API
2025 - Completed
NLP
Sentiment Analysis
Hugging Face
Transformer
RoBERTa
Streamlit

Mini 3: Sentiment Analysis API
2025 - Completed
NLP
Sentiment Analysis
Hugging Face
Transformer
RoBERTa
Streamlit
Mini 1: OCR Text Extractor with GCP Vision API & Streamlit
Overview:
Developed a Python tool with an interactive Streamlit web interface to extract text from images using the powerful Google Cloud Vision API.
Project Description
This project is a practical demonstration of leveraging cloud AI services for document processing. It provides a user-friendly web application where users can upload an image (JPG or PNG) and instantly receive the extracted text. The application also offers a feature to download the extracted text as a structured CSV file, bridging the gap between raw images and usable data.
Problem Statement
The process of manually transcribing text from images or scanned documents is tedious, time-consuming, and prone to error. While powerful Optical Character Recognition (OCR) technology exists, it often requires complex setup or integration. This project aimed to solve that problem by creating a simple, accessible, and highly accurate OCR tool that abstracts the complexity of a cloud-based AI service into a straightforward web interface, enabling anyone to digitize documents with just a few clicks.

Methodologies & Project Workflow
The project was executed in five distinct phases to ensure a structured and systematic development process:
Phase 1: Local Environment & Foundational Setup:
The initial phase focused on creating a standardized and isolated local development environment. This involved setting up a Python virtual environment for dependency management, initializing a Git repository for version control, and installing the core
google-cloud-visionlibrary.Phase 2: Google Cloud Platform Integration:
This phase involved configuring the necessary GCP services. The Cloud Vision API was enabled, and the local application was securely authenticated using Application Default Credentials (ADC) via the
gcloudCLI, eliminating the need to manage service account keys in the code.Phase 3: Core OCR Script Implementation:
A core Python script (
ocr_tool.py) was developed to handle the text extraction logic. This script used thegoogle-cloud-visionclient library to send image content to the API and process the returned text annotation.Phase 4: Interactive Web Interface Development:
To enhance usability, an interactive web interface was built using Streamlit. This involved creating a file uploader for images, displaying the results, managing application state, and implementing a feature to download the extracted text as a CSV file.
Phase 5: Final Documentation & Publishing:
The final phase involved creating a comprehensive
README.mdfile and publishing the complete project to a public GitHub repository. This ensured the project was well-documented and accessible as a portfolio piece.
Key Results
Project-Specific Outcomes: Developed a fully functional Streamlit web app for real-time OCR on uploaded images (PNG/JPG/JPEG) using Google Cloud Vision API's document_text_detection, enabling seamless text extraction, display, and download as .txt or .csv files—demonstrating end-to-end integration in a rapid 1-day prototype.
Efficiency and Usability: Achieved responsive performance via @st.cache_resource for API client caching, with live processing (spinners/alerts) completing in under a minute per image on local runs at http://localhost:8501.
Key Achievements: Resolved authentication hurdles (Google Cloud SDK/ADC) and corporate proxy issues for local development, showcasing practical skills in GCP integration and secure API handling without custom model training.
Model Insights: Leveraged the pre-trained Vision API for high-fidelity document text detection; qualitative tests on samples (e.g., receipt1.jpg) produced accurate extractions, though no quantitative metrics (e.g., accuracy) were evaluated—aligning with the project's focus on prototyping over benchmarks.
Key Features & Technical Details
Cloud-Powered OCR: Leverages Google Cloud Vision API for highly accurate and reliable text extraction from various image types.
Dual Interface: Offers both a command-line script for automated processing and an interactive web app built with Streamlit for ease of use.
Interactive Web UI: Allows users to easily upload JPG/PNG images, view the uploaded image, and see the extracted text in a clean text area.
Data Export: Includes a one-click download button to save the extracted text lines as a structured CSV file, ready for analysis.
Secure Authentication: Utilizes Application Default Credentials (ADC) for a secure, keyless authentication method to GCP services during local development.
Technical Stack
Cloud Platform: Google Cloud Platform (GCP)
AI Service: Cloud Vision API
Programming Language: Python
Web Framework: Streamlit
Core Libraries:
google-cloud-vision,pandasVersion Control: Git, GitHub
Challenges & Solutions
Challenge: Ensuring secure and seamless authentication between a local application and GCP APIs without hardcoding credentials.
Solution: Implemented Application Default Credentials (ADC) by configuring the
gcloudSDK (gcloud auth application-default login). This provided a secure and standard authentication mechanism that the Google Client Library could automatically use.
Challenge: The core OCR function was designed to work with file paths, but Streamlit's file uploader provides an in-memory byte stream, creating an incompatibility.
Solution: Integrated Python's
tempfilemodule to temporarily store the uploaded image data to a file. This generated a temporary file path that was passed to the existing OCR function, effectively bridging the web interface with the backend logic without significant refactoring.




Future Enhancements & Improvement
PDF & Multi-Page Support: Extend functionality to process multi-page PDF documents, extracting text from each page.
Cloud Deployment: Deploy the Streamlit application to a managed service like GCP's Cloud Run or Streamlit Community Cloud for public access.
Layout and Table Recognition: Enhance the processing to not only extract text but also recognize and preserve document structure, such as tables and paragraphs.
More Works
(GQ® — 02)
©2024

Project 1: Real-Time Fraud Detection API
2025 - Completed
API Development
XGBoost
FastAPI
AWS Cloud
Serverless Deployment
SHAP

Project 2: Automated MLOps CI/CD Pipeline
2025 - Completed
MLOps
CI/CD
Docker
Spark
Serverless Deployment
AWS Cloud

Project 3: QA Bot with Multi-turn RAG for e-Commerce
2025 - Completed
RAG
Multilingual
FastAPI
Gemini API
Hugging Face
Pinecone

Mini 2: Serverless ETL Pipeline on GCP for CSV Automated Processing to BigQuery
2025 - Completed
Data Engineering
ETL
Serverless
GCP
BigQuery
Cloud Functions

Mini 3: Sentiment Analysis API
2025 - Completed
NLP
Sentiment Analysis
Hugging Face
Transformer
RoBERTa
Streamlit
Mini 1: OCR Text Extractor with GCP Vision API & Streamlit
Overview:
Developed a Python tool with an interactive Streamlit web interface to extract text from images using the powerful Google Cloud Vision API.
Project Description
This project is a practical demonstration of leveraging cloud AI services for document processing. It provides a user-friendly web application where users can upload an image (JPG or PNG) and instantly receive the extracted text. The application also offers a feature to download the extracted text as a structured CSV file, bridging the gap between raw images and usable data.
Problem Statement
The process of manually transcribing text from images or scanned documents is tedious, time-consuming, and prone to error. While powerful Optical Character Recognition (OCR) technology exists, it often requires complex setup or integration. This project aimed to solve that problem by creating a simple, accessible, and highly accurate OCR tool that abstracts the complexity of a cloud-based AI service into a straightforward web interface, enabling anyone to digitize documents with just a few clicks.

Methodologies & Project Workflow
The project was executed in five distinct phases to ensure a structured and systematic development process:
Phase 1: Local Environment & Foundational Setup:
The initial phase focused on creating a standardized and isolated local development environment. This involved setting up a Python virtual environment for dependency management, initializing a Git repository for version control, and installing the core
google-cloud-visionlibrary.Phase 2: Google Cloud Platform Integration:
This phase involved configuring the necessary GCP services. The Cloud Vision API was enabled, and the local application was securely authenticated using Application Default Credentials (ADC) via the
gcloudCLI, eliminating the need to manage service account keys in the code.Phase 3: Core OCR Script Implementation:
A core Python script (
ocr_tool.py) was developed to handle the text extraction logic. This script used thegoogle-cloud-visionclient library to send image content to the API and process the returned text annotation.Phase 4: Interactive Web Interface Development:
To enhance usability, an interactive web interface was built using Streamlit. This involved creating a file uploader for images, displaying the results, managing application state, and implementing a feature to download the extracted text as a CSV file.
Phase 5: Final Documentation & Publishing:
The final phase involved creating a comprehensive
README.mdfile and publishing the complete project to a public GitHub repository. This ensured the project was well-documented and accessible as a portfolio piece.
Key Results
Project-Specific Outcomes: Developed a fully functional Streamlit web app for real-time OCR on uploaded images (PNG/JPG/JPEG) using Google Cloud Vision API's document_text_detection, enabling seamless text extraction, display, and download as .txt or .csv files—demonstrating end-to-end integration in a rapid 1-day prototype.
Efficiency and Usability: Achieved responsive performance via @st.cache_resource for API client caching, with live processing (spinners/alerts) completing in under a minute per image on local runs at http://localhost:8501.
Key Achievements: Resolved authentication hurdles (Google Cloud SDK/ADC) and corporate proxy issues for local development, showcasing practical skills in GCP integration and secure API handling without custom model training.
Model Insights: Leveraged the pre-trained Vision API for high-fidelity document text detection; qualitative tests on samples (e.g., receipt1.jpg) produced accurate extractions, though no quantitative metrics (e.g., accuracy) were evaluated—aligning with the project's focus on prototyping over benchmarks.
Key Features & Technical Details
Cloud-Powered OCR: Leverages Google Cloud Vision API for highly accurate and reliable text extraction from various image types.
Dual Interface: Offers both a command-line script for automated processing and an interactive web app built with Streamlit for ease of use.
Interactive Web UI: Allows users to easily upload JPG/PNG images, view the uploaded image, and see the extracted text in a clean text area.
Data Export: Includes a one-click download button to save the extracted text lines as a structured CSV file, ready for analysis.
Secure Authentication: Utilizes Application Default Credentials (ADC) for a secure, keyless authentication method to GCP services during local development.
Technical Stack
Cloud Platform: Google Cloud Platform (GCP)
AI Service: Cloud Vision API
Programming Language: Python
Web Framework: Streamlit
Core Libraries:
google-cloud-vision,pandasVersion Control: Git, GitHub
Challenges & Solutions
Challenge: Ensuring secure and seamless authentication between a local application and GCP APIs without hardcoding credentials.
Solution: Implemented Application Default Credentials (ADC) by configuring the
gcloudSDK (gcloud auth application-default login). This provided a secure and standard authentication mechanism that the Google Client Library could automatically use.
Challenge: The core OCR function was designed to work with file paths, but Streamlit's file uploader provides an in-memory byte stream, creating an incompatibility.
Solution: Integrated Python's
tempfilemodule to temporarily store the uploaded image data to a file. This generated a temporary file path that was passed to the existing OCR function, effectively bridging the web interface with the backend logic without significant refactoring.




Future Enhancements & Improvement
PDF & Multi-Page Support: Extend functionality to process multi-page PDF documents, extracting text from each page.
Cloud Deployment: Deploy the Streamlit application to a managed service like GCP's Cloud Run or Streamlit Community Cloud for public access.
Layout and Table Recognition: Enhance the processing to not only extract text but also recognize and preserve document structure, such as tables and paragraphs.
More Works
©2024

Project 1: Real-Time Fraud Detection API
2025 - Completed
API Development
XGBoost
FastAPI
AWS Cloud
Serverless Deployment
SHAP

Project 2: Automated MLOps CI/CD Pipeline
2025 - Completed
MLOps
CI/CD
Docker
Spark
Serverless Deployment
AWS Cloud

Project 3: QA Bot with Multi-turn RAG for e-Commerce
2025 - Completed
RAG
Multilingual
FastAPI
Gemini API
Hugging Face
Pinecone

Mini 2: Serverless ETL Pipeline on GCP for CSV Automated Processing to BigQuery
2025 - Completed
Data Engineering
ETL
Serverless
GCP
BigQuery
Cloud Functions

Mini 3: Sentiment Analysis API
2025 - Completed
NLP
Sentiment Analysis
Hugging Face
Transformer
RoBERTa
Streamlit