About Dura Digital
Dura Digital is a global digital transformation and consulting company that helps organizations turn emerging technologies into practical business outcomes. We bring together strategy, human-centered design, engineering, and AI expertise to create solutions that improve operations, customer experiences, and organizational performance. As a Dura Digital consultant, you will represent our commitment to thoughtful innovation, strong client partnership, and high-quality delivery while working closely with the client’s leadership and delivery teams.
About the role
We're hiring an AI Engineer to design, build, and ship AI-powered features that go into production and stay there. This is a hands-on engineering role, not a research role. You'll work across the full path from problem framing to deployment: turning an ambiguous business need into a working system, choosing the right model and retrieval approach, building the evaluation harness that proves it works, and standing behind it once real users depend on it.
You'll partner closely with product, design, data, and platform engineers, and depending on the engagement, directly with client stakeholders who are new to AI and need a clear-eyed guide on what's realistic.
What you'll do
- Build and deploy production AI features: LLM-backed workflows, retrieval-augmented generation, agentic and tool-using systems, classification and extraction pipelines, and traditional ML models where they're the better fit.
- Own the data path behind those features: ingestion, chunking, embedding, indexing, and retrieval quality, and improve it based on measured outcomes rather than intuition.
- Design evaluation before you ship. Define what "good" means for each use case, build offline eval sets and online feedback loops, and track regressions as prompts, models, and data change.
- Integrate AI capabilities into existing applications and services through well-designed APIs, with attention to latency, cost per request, failure modes, and graceful degradation.
- Instrument and monitor deployed systems for quality drift, hallucination rates, token spend, and abuse, and act on what you see.
- Apply responsible-AI practices in day-to-day work: data privacy and residency, PII handling, prompt injection and data exfiltration risks, access control, and clear documentation of model limitations.
- Prototype quickly to reduce uncertainty, then make a deliberate call on what deserves to be hardened and what should be thrown away.
- Contribute to internal standards, reusable components, reference architectures, and patterns other teams can adopt.
- Explain technical tradeoffs to non-technical audiences, including client stakeholders and executives, without overselling what the technology can do.
What you'll bring
- 3+ years of professional software engineering experience, including 1+ years building with LLMs or ML systems in production.
- Strong Python. Working proficiency in at least one of TypeScript/JavaScript, C#, Java, or Go for service and application integration.
- Demonstrated experience shipping an AI feature end to end, you can walk us through what you built, how you evaluated it, what broke, and what you changed.
- Practical command of the modern LLM toolkit: prompt design, structured output, function/tool calling, context management, and at least one orchestration framework or a well-reasoned argument for building without one.
- Experience with retrieval systems and vector search (e.g., pgvector, Azure AI Search, Pinecone, Elasticsearch, or similar), including how to diagnose bad retrieval.
- Cloud deployment experience on [Azure / AWS / GCP], plus containers, CI/CD, and infrastructure-as-code fundamentals.
- Solid data engineering instincts: SQL, working with messy real-world data, and building pipelines that don't silently fail.
- Clear written and verbal communication, and comfort operating with incomplete requirements.
Nice to have
- Consulting, agency, or client-facing delivery experience.
- Fine-tuning, distillation, or parameter-efficient adaptation of open-weight models, and a sense for when it's worth it versus prompting or retrieval.
- Experience with agent frameworks, MCP, or multi-step tool-using systems in production.
- MLOps tooling (experiment tracking, model registries, feature stores) and observability platforms for LLM applications.
- Background in a regulated domain such as healthcare, financial services, insurance, or the public sector.
- Speech, vision, or document-understanding work (OCR, layout parsing, multimodal models).
- Open-source contributions, technical writing, or conference speaking.