Senior AI Engineer

Job description

**Ataccama** is the agentic data trust company. Organizations worldwide rely on Ataccama ONE, the agentic data trust platform, to ensure data is accurate, accessible, and trusted for every decision and system. Powered by the ONE AI Agent, Ataccama ONE brings autonomy to data quality and governance, continuously monitoring, improving, and explaining the reliability of data across complex enterprise and multi-agent environments. At the core of the platform is Ataccama’s market-leading data quality, with modules built around it to unify data quality monitoring, catalog, lineage, observability, and reference data management, ensuring data is consistent and explainable. This quality-first foundation makes data the engine of trust, powering AI, analytics, and operations with confidence. Recognized as a **Leader in the 2026 Gartner Magic Quadrant for Augmented Data Quality and positioned furthest in Completeness of Vision**, Ataccama continues to set the standard for enterprise-grade data trust. Our people are located across the globe. They succeed by collaborating as a team and thrive in our company culture defined by these core values: **Challenging Fun** **ONE Team** **Customer Centric** **Candid and Caring** **Aim High** Our **AI team** are the evangelists of Artificial Intelligence at Ataccama. We build the AI agents that power data quality, cataloging, and governance workflows across our platform — moving cutting-edge research into reliable, production-grade features our customers depend on. As a **Senior AI Engineer**, you'll own AI-powered services end to end — design, delivery, and life in production. Much of what we build is agentic, so you'll be working on the LLM agents at the heart of our platform, whether you arrive having built them already or ready to master that here. It's a hands-on individual-contributor role, and we work you-build-it-you-run-it: whoever designs a service owns its reliability, its cost, and its pager. ### Your work will include: - Designing, building, and hardening LLM agents with real tool use and multi-step orchestration for data quality, catalog, and governance tasks. - Owning agent reliability end to end — latency, correctness, and cost — including evaluation harnesses and production observability. - Designing and evolving our agent tooling layer, including MCP servers and protocol-level integrations, so capabilities can be exposed cleanly and securely. - Building and tuning RAG pipelines and grounding strategies where agents need reliable access to customer data and metadata. - Working across our LLM gateway (LiteLLM proxy) and model providers to route, monitor, and optimize inference. - Setting technical standards for how we build agents, and raising the bar for the whole team through design reviews and pragmatic architectural trade-offs. ### What we're looking for - Strong, modern Python, and the engineering judgment of someone who has been senior for a while — you can be trusted with an ambiguous problem and a production system. - A track record of delivering a product or service end to end: you scoped it, built it, shipped it to real users, and kept it running. Not just features inside someone else's architecture. - A you-build-it-you-run-it mindset by conviction, not obligation — you've been on call for your own code, you've written the runbook, you've done the postmortem. - Solid grasp of production fundamentals: observability, testing strategy, CI/CD, incident response, and the operational cost of design decisions. - Comfort designing and operating services in cloud and Kubernetes environments. - Experience working with LLM APIs in production code (any provider) — you don't need to have built agents, but you should have shipped something backed by a model and dealt with what goes wrong. - Genuine engagement with where AI engineering is going — you follow how the field is moving and can reason about what's real versus noise. You've either built agents already, or you're set on mastering it here: this team's work is agentic, and we expect you to get there fast. - Healthy skepticism toward hype — you surface trade-offs before recommending a direction, and you say so when the simpler option is the right one. - Candid, low-ego collaboration with product, design, and other engineers. ### Nice to have - Experience building LLM agents that ran in production — tool calling, multi-step orchestration, and the failure modes that come with them. - Practical experience with MCP (or comparable protocol/tooling design), and the judgment to know when a protocol layer is worth it. - RAG, retrieval, and grounding techniques, and a realistic view of their limitations. - LLM gateways and proxies (e.g. LiteLLM) and provider APIs (Azure AI Foundry, OpenAI, or similar). - Agent-specific evaluation and quality measurement. - Background in data management, data quality, or governance. - OAuth2/SSO and multi-tenant service design.