Back to Projects

ResearchHub - AI Research Assistant

ResearchHub - AI Research Assistant - 1
Click to expand
ResearchHub - AI Research Assistant - 2
Click to expand
ResearchHub - AI Research Assistant - 3
Click to expand
Category:AI/ML Web Application
Client:ResearchHub (Personal Project)
Duration:January 2026 - Present
Year:2026

My Approach:
Crafting Digital
Excellence

View Live Project
View on GitHub

ResearchHub is an enterprise-grade AI research assistant inspired by Google's NotebookLM, designed to revolutionize how researchers interact with academic documents. This open-source platform combines cutting-edge AI technologies with an intuitive interface to provide a comprehensive research workflow solution.

Key Features

Neural Search Engine

  • Hybrid search combining BM25 lexical matching with vector embeddings.
  • AI-powered re-ranking for maximum relevance.
  • Real-time highlighted search results with confidence scores.

Intelligent Chat Interface

  • Streaming AI responses powered by Llama 3.3 70B.
  • Automatic citation tracking with page numbers.
  • Persistent conversation history across sessions.
  • Multi-document context awareness.

AI Podcast Generation

  • Transform research papers into conversational audio content.
  • Multi-voice synthesis with natural dialogue flow.
  • Synchronized transcript for easy navigation.

Knowledge Graph Visualization

  • Interactive 3D visualization of document relationships.
  • Automatic entity and concept extraction.
  • Semantic clustering for research discovery.

Intelligence Panel

  • AI-generated executive summaries.
  • Key insights and methodology analysis.
  • Research gap identification.

Export Capabilities

  • Multiple citation formats (BibTeX, APA, MLA, Chicago, IEEE).
  • PDF report generation.
  • Batch export functionality.

Technical Architecture

Frontend:

  • Next.js 14 with App Router.
  • TypeScript for type safety.
  • Tailwind CSS with custom design system.
  • Framer Motion animations.
  • Zustand for state management.

Backend:

  • FastAPI for high-performance API.
  • SQLAlchemy with async support.
  • ChromaDB for vector storage.
  • sentence-transformers for embeddings.

AI/ML Stack:

  • Groq API with Llama 3.3 70B.
  • RAG (Retrieval-Augmented Generation).
  • Edge TTS for audio synthesis.

Impact & Learning

This project deepened my expertise in:

  • Building production-grade RAG systems.
  • Implementing hybrid search algorithms.
  • Designing responsive, accessible UIs.
  • Managing complex state in React applications.
  • Optimizing AI model inference pipelines.

Future Roadmap

  • Multi-language support.
  • Collaborative research spaces.
  • API for third-party integrations.
  • Mobile application.

Other Projects

CrowdVision: AI-Powered Crowd Analytics
View Project →
PythonFastAPI PyTorchComputer VisionDockerFly.ioDeep Learning

CrowdVision: AI-Powered Crowd Analytics

An advanced AI web application capable of estimating crowd sizes with high accuracy. Utilizing dual deep learning models (CSRNet & P2PNet) to generate density heatmaps and precise point localization for real-time crowd analytics.

FashionAI - Clothing Recognition with Color Classification
View Project →
PythonComputer VisionYolov8FastAPIReal-time Detection

FashionAI - Clothing Recognition with Color Classification

FashionAI is an AI-powered real-time clothing detection system built with a dual-stage pipeline (YOLOv8 + Color Classifier) and FastAPI. It enables instant identification of 8 clothing categories through webcam live detection or image upload, featuring a premium dark-themed web interface and robust deployment ready for multi-platform environments.

Chess AI with Deep Reinforcement Learning
View Project →
Reinforcement LearningPyTorchDeep LearningChessPythonNeural NetworksFlaskStockfishHugging FaceAdaptive OptimizationPPO

Chess AI with Deep Reinforcement Learning

A neural network-based chess engine built with PPO and adaptive optimization, trained through 27 iterative versions to achieve strategic play against Stockfish.

Leksara
View Project →
PythonNLPE-commerceOpen-sourceData Science

Leksara

Leksara is an open-source Python toolkit for processing Indonesian text in the e-commerce domain, automating tasks like text cleaning, stopword removal, slang normalization, and punctuation handling. It helps Data Scientists and ML Engineers streamline their workflows and ensure clean, usable data for analysis.