AI Workflow Engineering Architect

About AppliedAI

Founded in 2021 and based in Abu Dhabi, UAE, AppliedAI is a trailblazer in artificial intelligence solutions for regulated sectors including healthcare, insurance, government, and financial services. Their flagship platform, Opus, automates and oversees mission-critical, documentation-intensive workflows, embedding governance, audit trails, and human supervision to drive productivity while enhancing reliability and trust.

Position Summary

The AI Workflow Engineering Architect serves as the lead technical expert responsible for designing and optimizing the production workflows within the Opus platform. This role focuses especially on machine learning-heavy and computation-intensive business processes. The architect defines workflow engineering standards to balance accuracy, latency, cost, and reliability by leveraging Opus capabilities and supports systems. The role also involves guiding Workflow Engineers through technical leadership, training, architecture assessments, reference engineering, and performance analysis.

Key Responsibilities

• Lead the breakdown of business processes into efficient Opus workflow graphs, outlining execution boundaries, dependencies, state handling, concurrency, and failure paths.
• Frame workflow design challenges as constrained optimization problems addressing accuracy, latency, throughput, inference cost, and operational reliability to determine balanced operating points.
• Set technical guidelines for Workflow Engineers selecting and composing machine learning models within Opus.
• Design and review diverse execution methods such as deterministic computation, specialized models, retrieval, model cascades, conditional routing, early exits, and human intervention.
• Ensure model selection and routing rely on empirical error metrics, confidence calibration, workload characteristics, and marginal inference economics.
• Establish engineering standards to validate Opus workflows, including defining representative test datasets, weighting loss functions by business impact, creating component and end-to-end benchmarks, and applying ablation, confidence interval, and regression threshold analysis.
• Guarantee evaluations address distribution shifts, uncommon cases, correlated failures, and costly errors rather than depending on average accuracy or limited test scenarios.
• Lead technical analysis of execution costs and performance aspects, covering inference, context building, retrieval, serialization, networking, concurrency, and orchestration overhead.
• Develop profiling and enhancement techniques for caching, batching, parallelization, model optimization (size, quantization), context lengths, and hardware accelerator utilization.
• Mandate performance characterization under realistic concurrency metrics including throughput, cost per successful execution, and latency percentiles (p50/p95/p99).
• Define best practices to ensure dependable workflow execution within Opus, emphasizing typed interfaces, explicit state transitions, idempotency, checkpointing, bounded retries, timeouts, backpressure, compensation, and partial failure recovery.
• Apply advanced techniques such as property-based testing, fault injection, deterministic replay, and trace analysis to assess reliability.
• Promote architectural understanding of failure modes amongst Workflow Engineers, treating failures as design properties, not post-deployment issues.
• Specify how to measure workflow performance post-deployment and how production data informs ongoing engineering decisions.
• Implement instrumentation to correlate workflow versions, model and configuration choices, execution logs, errors, latency, inference resource usage, and business outcomes.
• Lead root cause analysis for performance regressions and determine appropriate corrective actions across workflow structure, data, model selection, routing, context, implementation, or capacity.
• Mentor Workflow Engineers to develop independent expertise in machine learning and computer science decision-making.
• Spearhead complex workflow designs, technical reviews, post-mortems, maintain standard references and reusable patterns in Opus.
• Act as the principal technical authority on workflow engineering to transform expert knowledge into a consistent engineering discipline throughout AppliedAI.

Required Qualifications

• Extensive experience in applied machine learning and software engineering with practical involvement in production deployment of large language models or agentic AI systems (not limited to research or prototypes).
• Strong expertise in machine learning evaluation methodologies and system performance metrics such as latency, cost, and concurrency, treating these as core engineering concerns.
• Fundamental knowledge of distributed systems emphasizing reliability, fault tolerance, and state management in production pipelines.
• Proven track record of mentoring engineering teams and establishing technical standards beyond individual contributions.

Preferred Qualifications

• Experience orchestrating multiple models using routing, cascades, or fallback strategies across providers such as Anthropic, OpenAI, and Gemini.
• Background working in regulated sectors like healthcare, finance, insurance, or business process outsourcing.
• Demonstrated ability to create engineering standards or technical disciplines from the ground up within startup or early-stage companies.
• Published research, patents, or documentation related to machine learning systems or optimization of inference processes.

Benefits

• Play a critical role in shaping marketing analytics strategies at a forward-looking AI enterprise.
• Gain experience with state-of-the-art AI/ML applications in practical, real-world business situations.
• Collaborative and innovation-driven environment that encourages professional growth.
• Competitive pay, benefits package, and opportunities for career progression.