Applied Scientist

Title: Applied Scientist

Work Location: On-site, Los Angeles, CA

Start Date: ASAP, subject to candidate availability

Technology Environment: AWS technology stack, working closely with the AWS team

Interview Process: 2 rounds

Engagement: Enterprise, revenue-critical real-time streaming/AI program

Role Overview: Design, build and optimise GenAI-driven multimodal pipelines for detecting contextual moments in live video streams. The role focuses on prompt engineering, multimodal inference and model optimisation using foundation models such as Claude, Nova and Amazon Bedrock to maximise detection accuracy.

Key Responsibilities

• Design and optimise prompt engineering strategies for foundation models
• Build pipelines for multimodal inference across:
• Video
• Transcript
• Audio
• Detect and classify custom moments and contextual events
• Develop reusable frameworks for:
• Moment detection templates
• Structured metadata generation
• Prompt optimisation
• Fine-tune prompts and configurations to improve:
• Accuracy
• Precision across different moment types
• Integrate GenAI outputs into:
• Structured JSON metadata
• Downstream systems
• Data processing pipelines
• Work with AWS services including:
• Amazon Bedrock
• AWS Lambda
• Amazon S3
• Amazon Kinesis
• Collaborate with the QA team for output validation and optimisation
• Support production rollout and continuous improvement

Required Skills & Experience

Strong experience in:

• Prompt engineering for foundation models such as Claude, Nova, etc.
• Multimodal AI systems covering vision and NLP
• LLM inference and optimisation
• Structured outputs, including JSON generation and validation
• Strong Python programming and data pipeline development
• AWS cloud stack, particularly:
• Bedrock
• Lambda
• S3
• Analytics/data pipelines
• Video analytics or media AI workflows
• Real-time or near-real-time AI systems