CSS-1:3RDi_AI/MLEngineer

About the Role

We are seeking a highly skilled and motivated Generative AI Engineer with hands-on experience in building and deploying multimodal AI solutions (text, audio, and video) on Azure or AWS AI platforms. The ideal candidate will have deep knowledge of generative AI models (LLMs, diffusion models, etc.), and be capable of designing, developing, and scaling AI/ML solutions using cloud-native tools and services.

Key Responsibilities

• Design, develop, and deploy multimodal AI models (text, audio, video) using AWS or Azure AI services.
• Integrate LLMs (e.g., OpenAI, Anthropic, Cohere) with proprietary or third-party models to build custom applications.
• Utilize Azure OpenAI, AWS Bedrock, SageMaker, or similar services for training and inference workflows.
• Develop pipelines for preprocessing, training, evaluation, and fine-tuning of multimodal models.
• Collaborate with product and engineering teams to define use cases and translate them into scalable GenAI solutions.
• Evaluate and optimize models for latency, accuracy, cost, and ethical considerations.
• Implement prompt engineering, RAG (retrieval-augmented generation), vector search, and other advanced techniques.
• Stay current with GenAI research trends and contribute to internal knowledge sharing.

Required Skills & Qualifications

• 5+ years of experience in AI/ML, with 1 2 years specifically in generative AI / multimodal AI.
• Strong proficiency with at least one cloud platform: Azure AI (Azure OpenAI, Cognitive Services, ML Studio) or AWS AI (Bedrock, SageMaker, Transcribe, Rekognition, Polly, etc.).
• Hands-on experience with multimodal data processing (e.g., NLP, audio signal processing, video analysis).
• Experience with LLMs, transformer architectures, and model fine-tuning (e.g., LoRA, PEFT, adapters).
• Proficiency in Python, and libraries such as PyTorch, TensorFlow, Transformers (Hugging Face).
• Experience with vector databases (e.g., FAISS, Pinecone, Weaviate) and embedding generation.
• Strong understanding of MLOps, CI/CD for ML, and deployment strategies for GenAI models.