The Data Dilemma

Open Source Intelligence Meets Real-World Impact

Transforming healthcare with ethical AI solutions that prioritize privacy, accessibility, and community collaboration.

Open-Source AI Tools

Code-Switched Speech Processing

Open-source tools that handle multilingual audio where speakers switch between languages mid-conversation—a common challenge in healthcare settings with diverse patient populations.

Privacy-Preserving Healthcare AI

Community-built solutions that maintain patient confidentiality while enabling powerful AI analysis, designed specifically for healthcare's strict privacy requirements.

Low-Resource Language Support

Collaborative tools that work effectively with limited training data, making AI accessible for underserved languages and communities in healthcare contexts.

Noise-Robust Audio Processing

Open frameworks that clean and process audio data from real-world healthcare environments—handling background noise, poor recording quality, and challenging acoustic conditions.

Deployable Healthcare Models

Production-ready, open-source AI models that bridge the gap between research and real clinical applications, with comprehensive documentation and community support.

Our GitHub Projects

MediBeng Whisper Tiny

Fine-tuned ASR model for Bengali-English speech translation. View on GitHub

LLM Application

A real-time AI-powered chat app using FastAPI, WebSocket, and Groq. View on GitHub

MediRag Guard

A RAG delivers comprehensive, context-aware insights on healthcare data privacy through a novel knowledge tree. View on GitHub

Medibeng-Orpheus-3b-0.1-ft-Fine-Tuning

A fine-tuned Text-to-Speech (TTS) model trained on the MediBeng dataset, designed to handle bilingual Bengali-English code-switching in healthcare settings. View on GitHub

Our Research Contributions

Recognition of Sunflower Diseases Using Hybrid Deep Learning

A hybrid AI model combining transfer learning (VGG19) and CNN for efficient sunflower disease detection, achieving superior precision, recall, and accuracy compared to other approaches. Read Paper

MediBeng Whisper Tiny: Clinical Code-Switched Translation

We present the MediBeng Whisper Tiny model, a cost-effective, fine-tuned solution for accurate transcription and translation of code-switched Bengali-English conversations in healthcare. Read Paper

COVID-19 Fake News Detection

This study analyzes machine learning techniques, with the Random Forest ensemble method achieving 90% accuracy in identifying fake news related to the COVID-19 pandemic. Read Paper

Join Our Open-Source Mission

Code & Development

Contribute to our GitHub repositories, help improve existing tools, or build new solutions for healthcare AI challenges. All skill levels welcome—from bug fixes to major feature development.

Research & Testing

Share your healthcare data challenges, test our tools in real-world scenarios, or collaborate on research papers. Your domain expertise helps make our solutions more practical and effective.

Documentation & Community

Help others by improving documentation, creating tutorials, answering questions, or sharing use cases. Building a supportive community is just as important as building great code.

Partnership & Funding

We welcome mission-aligned partnerships, grants, and collaborations that advance ethical healthcare AI. We also offer implementation consulting to help organizations deploy our open-source tools effectively.

Ideas & Feedback

Have a healthcare data problem that needs solving? Want to suggest new features or research directions? Your insights from the field help guide our development priorities and real-world impact.

Contact Us

For inquiries, collaborations, or any questions, feel free to reach out to us.

Email: me@promila.info

Address: Nirala, Khulna, Bangladesh

Let's talk! Email us or connect on LinkedIn.