04 Subject 1 : Real-Time AI Co-Influencer for Live Social Media Growth PFE

Project overview

• Design and implement a Real-Time AI Co-Influencer system that assists and augments live social media broadcasts to drive audience growth and engagement.
• Focus on low-latency multimodal AI (audio, video, chat) that operates alongside a human host to recommend actions, generate content, and interact with viewers in real time.

Objectives & expected deliverables

• Deliver a working prototype capable of real-time inference during a live stream: speech-to-text, intent classification, suggestion generation, and short-form content suggestions (comments/captions).
• Provide quantitative evaluation showing improvement in growth/engagement metrics (e.g., viewer retention, chat activity, likes, follows) from controlled tests or simulated sessions.

Core tasks and technical scope

• Implement low-latency ML pipelines (speech recognition, NLP for intent/response generation, lightweight vision models for scene/context detection).
• Integrate with streaming protocols and social media APIs (RTMP/WebRTC, chat APIs) to inject suggestions and moderate/automate interactions.
• Optimize models for real-time deployment (quantization, ONNX or TensorRT export) and package components with containerization for reproducible runs.

Technologies & tools (examples)

• Python, PyTorch or TensorFlow for model development; ONNX/TensorRT for inference optimization.
• WebRTC/RTMP, REST APIs for social platforms, and lightweight front-end/ui for presenter feedback.
• Docker for packaging, and optional cloud/edge deployment for low-latency inference.

Required profile and skills

• Strong Python skills and practical experience in ML (NLP and/or CV) and real-time inference concerns (latency, batching, model size).
• Experience with speech-to-text, transformer-based text generation or dialogue systems, and integrating with web/streaming APIs.
• Good software engineering practices: version control, containerization, clear experiments and evaluation reports.

Supervision, milestones and evaluation

• Milestones include: literature/proof-of-concept, prototype integration with live stream, optimization and deployment, and evaluation on engagement metrics.
• Final submission should include source code, deployment instructions, a short demo video of the system in a live or simulated stream, and an evaluation report with metrics.

How to apply

• To apply for this specific PFE project, send your CV and a brief motivation mentioning this project title to career@olindias.com.
• Use the project title in email and provide links to relevant code or demos if available.