Job details
About this role
Role overview This is a short, fixed-scope proof-of-concept engagement for an AI Engineer who can turn model-integration concepts into a working serverless layer on AWS. The mission is to build a thin proxy that wraps multiple third-party video and image generation models behind a single AWS Lambda and API Gateway surface, then validate it end-to-end with a frontend team that runs on Firebase and GCP. The work is delivery-oriented and runs roughly 15–20 business days, so the right candidate is hands-on, autonomous, and comfortable defending technical trade-offs directly with stakeholders.
Responsibilities
- Architect and deploy a serverless proxy layer using AWS Lambda fronted by API Gateway, provisioned through Infrastructure-as-Code (AWS CDK or CloudFormation), mirroring the existing model-call pattern without changing the client's Firebase deployment pipeline. - Extend six model integrations (Veo, Sora, SeedDance for video; GPT-Image, Nano Banana, FLUX for image) to accept and forward a 9:16 vertical aspect ratio parameter, preserving existing prompt, style, and camera controls. - Build a direct REST integration against ByteDance/ModelArk with bearer-token authentication, wrapped in its own Lambda function and exposed through the existing UI dropdown pattern. - Implement a content-moderation pre-check using Amazon Bedrock ahead of FLUX and SeedDance calls, screening for celebrity or political likeness and graphic content, with the final mechanism (Guardrails policy, LLM classification, or a combination) decided during design. - Validate the full flow through a new demo project that exercises story ideation, storyboard image generation, 9:16 SeedDance video generation, and MP4 export in a Beta environment. - Document key architectural decisions, deployment steps, and integration patterns for handover to the client's engineering team.
Requirements
- 3+ years of hands-on experience as a backend or AI engineer with a strong software-engineering foundation, not just data-science experimentation. - Production-level AWS experience, including Lambda, API Gateway, IAM, and reproducible provisioning via AWS CDK or CloudFormation. - Familiarity with Amazon Bedrock, particularly Bedrock Guardrails or LLM-based classification for content moderation, with the ability to ramp quickly if direct exposure is limited. - Demonstrated skill integrating third-party REST APIs with bearer-token authentication into clean, production-ready service layers. - Full-stack awareness to collaborate with an existing React and Firebase frontend without owning UI design or frontend development. - Delivery-oriented mindset, able to operate independently inside a short, fixed-scope engagement and communicate trade-offs clearly in working English with the client team.