software engineer @motive • ex backend & ai intern @omnisynkai • backend dev • system design • AI enthusiast

projects

github

Sachin Dapkara  • Peerlist

rubberduck-mcpMake AI coding feel human.rubberduck-mcp is an open-source MCP server that transforms your AI assistant into a real-time collaborator. It introduces two lightweight but powerful tools: clarify() and yap().With clarify(), the AI pauses and asks you when it’s confused - no more mysterious decisions or silent mistakes. With yap(), it shares its thoughts and reasoning as it works - like a helpful, slightly dramatic teammate who always keeps you in the loop.<h3>What it looks like in action:</h3>clarify() "Should I use Flexbox or Grid to center this form? There are like 7 ways to do it — any preference?"yap() "Hmm, this endpoint’s returning 500. Might be the DB... Nope, bcrypt’s acting up again. Let me fix that."You don’t just see the code - you see what your AI is thinking, struggling with, and trying to solve.<h3>Why it matters:</h3><ul><li>AI asks for help instead of silently guessing</li><li>You get real-time insights into its decisions</li><li>Feels more like pair programming, less like prompt engineering</li><li>Fewer mistakes, better communication, smoother flow</li></ul><h3>Designed for devs:</h3><ul><li>One-minute setup</li><li>Works with any MCP-compatible IDE</li><li>Zero lock-in, 100% open source</li></ul>npm install -g rubberduck-mcp GitHub: <a target="_blank" rel="noopener noreferrer" class="break-all link-break text-gray-500 hover:underline" href="https://github.com/superiorsd10/rubberduck-mcp">github.com/superiorsd10/rubberduck-mcp</a>Built to make AI feel like a teammate - not a black box.

rubberduck-mcp

Agrico helps organic farmers in collecting data and decide what is best for increasing their production yield.

Agrico

This project aims to develop a machine-learning model capable of distinguishing between human-generated and AI-generated text. By leveraging various linguistic, syntactic, semantic, and additional text features, the model aims to accurately identify the authorship of given text samples. The implementation involves data preprocessing, feature extraction, model training with hyperparameter tuning, and evaluation using a voting classifier that combines multiple machine learning algorithms.

AI vs Human Text Classification

This project presents a comprehensive solution for the Amazon Machine Learning Challenge, focusing on extracting product attributes from images and text through advanced techniques in natural language processing and computer vision. Key highlights include:

- Technical Architecture:
  - Developed a custom data pipeline using a "ProductDataset" class for efficient dataset handling.
  - Implemented asynchronous image downloading and flexible CSV parsing for improved performance.

- Model Implementation:
  - Utilized the "vikhyatk/moondream2" model for seamless integration of image and text inputs, enabling high-performance visual question answering capabilities.

- Inference Optimization:
  - Optimized the inference process for GPU acceleration and batch processing to ensure rapid predictions.

- Post-Processing Algorithms:
  - Designed sophisticated algorithms for accurately extracting various attributes, including:
    - Weight: Used regex for precise numerical value extraction.
    - Voltage: Handled multiple voltage units and formats.
    - Wattage: Processed complex representations and managed unit prefixes.
    - Dimensional Attributes: Parsed multi-dimensional formats effectively.
    - Volume: Managed compound units with intelligent defaulting.

- Challenges Addressed:
  - Overcame multi-modal input handling through a custom dataset class.
  - Optimized model deployment for improved inference speed.
  - Developed regex patterns for complex string parsing.
  - Established comprehensive unit mapping for consistency across predictions.
  - Implemented robust error handling to maintain pipeline stability.
  - Formulated adaptive questions based on target attributes for optimized model responses.

Amazon Machine Learning Challenge: Multi-Moda

Todo Applications might look easy to code and develop. But we present to you the most complicated Todo Application, which works normally and simply from the client side but has one of the most complex backends powered over polygon blockchain and Multiple databases.

Block-ToDo

College GateKeeper is an innovative security management application designed to streamline and enhance the monitoring of student attendance and security within educational institutions. This project empowers college security guards with a comprehensive toolset to manage access and track student movements effectively.

Key Features:

• Secure Authentication:
College GateKeeper ensures a secure environment with individual login credentials for each security guard. Additionally, a master password adds an extra layer of authentication to validate access requests.

• ID Card Image Capture:
Security guards can effortlessly capture students' ID card images using the app. This feature facilitates quick and accurate identification of students within the college premises.

• Roll Number Retrieval:
Upon capturing the ID card image, the app fetches the scanned roll number from the server. Guards have the option to confirm the information or retake the image for accuracy.

• Attendance Logging:
The application seamlessly records the out time when a student leaves the college premises. This information is securely stored in the database, providing a reliable record of student movements.

• In-Time Recording:
When students return to the college, their ID cards are scanned again to log their in-time. This ensures a comprehensive attendance tracking system for accurate reporting.

• Defaulter's List:
College GateKeeper automatically identifies students whose in-time exceeds the permitted limit. Details of such students are promptly added to the defaulter's list, facilitating the imposition of fines as necessary.

Overall, College GateKeeper not only enhances the efficiency of security operations but also contributes to maintaining discipline by enforcing attendance policies. Implementing this project demonstrates a commitment to leveraging technology for a secure and organized educational environment.

College GateKeeper

Confess helps reduce the burden your heart and mind are carrying. Because you are anonymous, you can confess anything without worrying about other people's thoughts. Only that confession will be displayed when you confess, and your identity will remain hidden.

Confess

Dash AI is an open-source coding assistant specifically designed for Flutter Devs. The assistant allows you to chat with Gemini inside VSCODE and create, refactor, and debug code.

Dash AI by Welltested-AI

EduHub-AI: Revolutionizing Online Education with AI-Powered Learning

EduHub-AI is a cutting-edge online learning platform that harnesses the power of artificial intelligence to automate tasks and enhance the educational experience. Designed to streamline the teaching and learning process, this comprehensive platform offers a range of innovative features that cater to the needs of both instructors and students.

Key Features:

- Live Streaming and Recordings: Instructors can conduct live-streamed classes, with recordings automatically captured and made available for students to access and review at their convenience.

- AI-Generated Assignments and Quizzes: Utilizing advanced language models like Llama3-70b, EduHub-AI can automatically generate assignments and quizzes, tailored to individual student performance and adaptive difficulty levels.

- Automated Grading and Feedback: Powered by prompt engineering techniques, the platform offers automated grading and personalized feedback on assignments, streamlining the evaluation process for instructors.

- Contextual Chat: Students can engage in contextual chat with lecture recordings and course materials (PDFs, PowerPoint presentations, Word documents), enhancing their understanding and facilitating collaborative learning.

- Plagiarism Checker: EduHub-AI incorporates a plagiarism-checking feature, ensuring academic integrity and promoting original work.

- Group Chat and Collaboration: Students and instructors can participate in group chats, fostering seamless communication and collaboration within educational hubs.

Technical Details:

- Tech Stack: Flask, Next.js, Celery, MongoDB, Redis, AWS, 100ms, Socket.IO

- Key Implementations: RAG architecture for contextual chat, webhook listener for transcript extraction, OpenCV and LLaVA-v1.6-34b for image frame processing, difflib module for plagiarism checking, and AWS services (S3, Lambda, FFmpeg, CloudFront) for file management and delivery.

EduHub-AI

Get Up is a simple alarm application where you can set an alarm by picking the time for yourself.

Get Up

Its Money is a simple fun application that predicts your salary after your placements based on your interest in Competitive Programming, and Development.

It's Money

Proctify is a cutting-edge platform designed to address the growing challenges of maintaining integrity in online tests and coding competitions. Whether it's hosting a third-party test like Google Forms or creating custom coding contests, Proctify ensures a secure, monitored, and seamless experience for both test creators and participants.  

Key Features:
- Advanced Proctoring: 
  - Detects violations such as absence from the camera, multiple faces, high audio levels, and prohibited objects (e.g., phones).  
  - Monitors browser activity, preventing tab changes, right-clicks, and copy-paste, while ensuring participants remain in fullscreen mode.  

- Versatile Test Hosting:  
  - Support for third-party linked tests by simply entering the test URL, along with details like start time and duration.  
  - Custom coding contests with problem creation, sample inputs/outputs, and real-time leaderboard functionality.  

- Real-Time Monitoring:  
  - Admins receive an Unfair Means (UFM) score for participants, highlighting potential cheating activities.  
  - Dashboards for creators to monitor tests, access detailed reports, and ensure fairness.  

Why Proctify Stands Out:
- Combines proctoring with the flexibility to host coding competitions and third-party tests.  
- Provides real-time violation detection and analytics, ensuring a fair environment for all participants.  
- Built with state-of-the-art technology, emphasizing scalability, performance, and user experience.  

With Proctify, I contributed to solving a major problem in the education and tech industry, showcasing my skills in system design, backend development, and cutting-edge technology integration.

proctify

This project is a secure pastebin-like application designed to allow users to create, manage, and access encrypted pastes. It provides user authentication, secure paste storage, and click tracking, with data managed across multiple services including AWS S3, Redis, and PostgreSQL.

Features

- Register and log in using a username and password.
- JWT token-based authentication for secure API access.

- Users can create pastes with metadata including title, content, tags, visibility (public/private), expiration time, and security settings.
- Unique ID generation for each paste using nanoid, validated for uniqueness with a Bloom filter.
- Content encryption with AES-256-CBC algorithm, and CRC32 checksum for integrity verification.
- Validation to ensure content size does not exceed 1MB.
- Paste content is stored securely in Amazon S3.

- Private pastes require authentication; access is controlled based on user permissions.
- Cached content retrieval from Redis to improve performance and reduce S3 calls.
- Metadata and content retrieval using signed URLs from S3 when not cached.
- Paste expiration and visibility checks, password protection, and data integrity verification.
- Secure pastes are decrypted upon retrieval if necessary.

- Increment click counts in Redis, and periodically update PostgreSQL with batch processing.

psst

Get bit by the nostalgia bug with Snake Bit! Choose from three modes, Overflow, Underflow, and Red Bit, and slither your way to high scores and endless fun.

Snake Bit

This project aims to build a non-conventional alarm clock with smart features such as auto-dismissal based on phone activity, weather, and more! It also includes challenges to ensure you completely wake up and an option to set shared alarms! This is the ULTIMATE alarm clock :)