AI Engineer Needed for Embeddings & Image-To-Text Optimisations

Mifu is an AI-powered influencer marketing platform that helps brands scale creator campaigns by automating influencer discovery and campaign management through AI. We are seeking an expert to take ownership of our existing AI data pipeline and enhance its performance, cost-efficiency, and recall quality. The initial priority will be to optimise our image-to-text and embeddings pipelines for speed and cost, improve batching, caching, concurrency, and queuing in a serverless AWS environment, and tune performance of models such as InternVL and Qwen via RunPod and Replicate. Ongoing work will include improving semantic recall and retrieval quality for influencer content using Pinecone and OpenSearch, and experimenting with multi-modal embeddings for better search accuracy. There is also scope to extend the system into RAG and agentic AI workflows for automated insights and recommendations. The ideal candidate will have hands-on experience with optimising AI/ML pipelines—especially image and embedding workflows—in AWS and Python, a strong grasp of multimodal embeddings and semantic search, and familiarity with LangChain, RAG, and vector databases.

Budget range:
Receive offers