AI Engineer (Fullstack+ GenAI)

Job description

Job Description: AI Engineer Job Description Location : Dentsu India Team : Digital Product, Pipes Level: 35 Team Description The AI Engineer role is placed within the UK Digital Product team, part of Dentsu's Data & Technology group. Digital Product is responsible for building and maintaining Dentsu's internal software products, built in partnership with some of the world's leading technology firms, including Meta, Google and Amazon among others. The Digital Product team is a hybrid team, with members split between the UK and India. The core product that you would be working on, Pipes, helps Dentsu teams optimise digital marketing campaigns for some of the largest advertisers in the world. Our platform is used by hundreds of users within the dentsu network to manage over 1,000 clients across 28 markets. Role Overview Over the last 6 months we have developed native agents within the Pipes platform, which are live in production, covering Google Ads, Microsoft Advertising, Meta, TikTok, etc. We're now scaling this significantly, and are hiring 2 AI Engineers on fixed-term contracts to help us get there. This role is about shipping GenAI agents end-to-end: integrating with ad platform APIs (via Model Context Protocol (MCP) client layer or direct API calls), defining the agent itself, authoring and iterating on prompts, and building the React frontend surface users interact with. You'll be assigned ownership of either our Search agents (Google Ads, Microsoft Advertising) or our Social agents (Meta, TikTok), and will be expected to ship 4-5 agents across the length of the contract, each with an accompanying automated test suite landing alongside it rather than after. You'll work day-to-day alongside a lead engineer, who owns architecture, cross-agent review and unblocking, a test engineer, who partners with you to build the eval suite for each agent you ship, and the teams who will be using the agent, who will advise on agent requirements and provide feedback. You're expected to work with a good degree of autonomy once onboarded, taking ownership of your agents from design through to production. Key Responsibilities Agent Delivery Build and ship new GenAI agents end-to-end: platform API/MCP integration, agent definition, prompt authoring, and the frontend chat experience. Own delivery of 4-5 agents across your assigned channel track (Search or Social), working largely independently once onboarded. Author and iterate on prompts using our prompt management tooling Collaborate with the test engineer so each agent ships with an automated eval/regression suite alongside it. Required Experience & Skills 3-5+ years' experience in a software engineering role. Strong experience with C#/.NET (or another object-oriented language) Experience with a JavaScript frontend framework (React preferred). Hands-on experience building GenAI/LLM-powered applications, including function/tool calling and prompt engineering (experience with any agent framework is transferable; we use Microsoft's Agent Framework). Experience integrating third-party or platform APIs. Ability to work independently, taking a feature from design through to production with light supervision. Experience working in a sprint-based development team. Desirable / Nice-to-Have Skills Experience with some of the following is beneficial but not required: Exposure to MCPs, or building tool-use integrations for LLMs. Experience with Google Vertex AI, Gemini, or a comparable LLM platform. Familiarity with ad platform APIs (Google Ads, Microsoft Advertising, Meta, TikTok, DV360). Experience with Firestore or Google Cloud Storage. Tech Environment The Pipes product is built using: Backend : C# (.NET 10), Microsoft Agent Framework, Google Vertex AI (Gemini) Frontend : ReactJS Data : MySQL, Firestore, Google Cloud Storage Infrastructure : Google Cloud Platform (Cloud Run, Cloud Functions, Cloud Scheduler, Docker) Prompt Management : PostHog Testing : xUnit (C#), Cypress (frontend) — general repo conventions; plus a new eval/golden-set framework for agent output, owned by a test engineer What Success Looks Like Within the contract period, success in this role will include: 4-5 production agents shipped for your assigned channel, each with platform integration, a working frontend experience, and an eval suite landed alongside it. Demonstrated ability to own agent delivery end-to-end with minimal day-to-day guidance. A track record of low defect/false-positive rates in your shipped agents' output reaching real users. Location: DGS India - Bengaluru - Manyata N1 Block Brand: Merkle Time Type: Full time Contract Type: Permanent