Senior Applied AI Engineer, Forward Deployed

About OryonTech

OryonTech develops Agent as a Service, a WhatsApp-first platform that enables businesses to manage customer interactions, orders, bookings, payments, and operational workflows. The platform combines conversational AI with business systems and human oversight.

Our engineering priorities are reliable execution, secure multi-tenant operation, and measurable value for the businesses using the product.

Position overview

We are seeking a senior engineer to take technical ownership of our AI agent systems and lead the delivery of capabilities from customer discovery through production operation.

This role combines applied AI engineering, backend development, and direct engagement with business users. You will shape how our agents interpret requests, maintain context, access business data, execute workflows, and transfer control to human operators. You will assess the existing platform, establish technical priorities, and implement improvements that support both current deployments and future product development.

You will remain directly involved in architecture, code, evaluation, and production troubleshooting. Working with backend, infrastructure, product, and GTM colleagues, you will translate customer requirements into solutions that can be maintained and reused across the platform.

The role carries responsibility for engineering decisions, delivery quality, and the operational performance of the capabilities you own.

AI-assisted engineering (required)

You must actively use company-approved AI coding assistants and agentic development tools as part of your daily engineering workflow to build and improve production software efficiently. This is a core requirement, not an optional advantage, and applies to your own development process, not only to building AI product features.

• Use AI tools for research, planning, prototyping, implementation, refactoring, testing, debugging, documentation, and production diagnosis when appropriate to the task.
• Supply relevant codebase context, API contracts, constraints, failure cases, and acceptance criteria; iterate deliberately rather than accepting the first generated answer.
• Understand, review, test, and explain every change you ship. AI output does not replace engineering judgment, code review, security checks, or deployment approval.
• Protect credentials, personal information, customer data, and proprietary code. Use only approved tools and data-handling settings, keep access least-privileged, and never give AI tools uncontrolled production access.
• Build reusable prompts, instructions, tests, and automation that reduce repetitive work. Demonstrate efficiency through delivery time, reduced rework, and reliable outcomes, not lines of generated code; keep tool usage and cost proportionate.
• Use AI tools to accelerate agent and tool implementation, evaluation harnesses, regression tests, retrieval experiments, and production failure analysis; validate agent actions against backend-confirmed state.

During technical screening, be ready to walk through an AI-assisted change, explain what you accepted or rejected, and show how you verified correctness. No additional document or exercise is required with your initial CV email.

Principal responsibilities

AI architecture and product development

• Lead the design and evolution of agent orchestration, tool interfaces, conversational state, memory, retrieval, and human handoff.
• Develop production Python services and LangGraph workflows. Evaluate architectural trade-offs and improve existing components before introducing additional complexity.
• Assess model capabilities and limitations; determine when hosted models, retrieval, fine-tuning, custom models, or deterministic software are appropriate.

Customer discovery and technical delivery

• Lead technical discovery with merchants and internal stakeholders. Translate business workflows into requirements, solution designs, acceptance criteria, and delivery plans.
• Own implementation, integration, evaluation, rollout, and post-deployment verification for assigned capabilities.
• Convert recurring deployment needs into configurable product features, reusable integrations, and documented implementation patterns.

Reliability, security, and operational control

• Engineer reliable handling of concurrent requests, duplicate events, retries, timeouts, interrupted workflows, and external-service failures.
• Maintain backend authority over permissions, pricing, inventory, payments, bookings, and fulfillment. Ensure agent responses reflect confirmed system state.
• Enforce tenant and session isolation, secure access to tools and data, and appropriate handling of sensitive information.
• Work with infrastructure engineers on GCP deployments, observability, incident response, controlled releases, and recovery procedures.

Evaluation and performance

• Establish representative evaluation datasets, regression suites, and release criteria for agent behavior and transaction correctness.
• Diagnose failures across models, orchestration, retrieval, tools, and backend services. Validate improvements against documented baselines.
• Measure task completion, latency, model usage, and operating cost. Use these findings to guide technical priorities and model selection.

Technical leadership

• Lead design discussions and code reviews, mentor colleagues, and establish maintainable implementation standards.
• Document material technical decisions, system dependencies, and operational procedures.
• Communicate delivery risks and trade-offs clearly. Ensure knowledge and operational responsibilities are shared across the team.

Required qualifications

• Demonstrated practical use of AI-assisted engineering tools to deliver tested software, with the ability to independently explain, debug, and maintain the resulting work.
• A substantial record of professional software engineering, with demonstrated senior-level responsibility for designing, delivering, and operating customer-facing systems.
• Direct experience deploying AI or LLM applications to production, with evidence of your contribution, evaluation approach, and operational results.
• Advanced Python proficiency and experience developing tested backend services with FastAPI or a comparable framework.
• Practical LangGraph experience covering orchestration, state persistence, tool execution, and failure diagnosis.
• Strong knowledge of relational data modeling, SQL, transactions, and data integrity, with PostgreSQL or comparable database experience.
• Experience with Redis, asynchronous processing, background jobs, API integrations, and event-driven systems.
• Hands-on GCP deployment and operational experience, supported by working knowledge of Docker, CI/CD, service permissions, secrets management, and monitoring.
• Sound understanding of LLM evaluation, retrieval, context management, and model limitations, including data leakage and the distinction between fluent responses and correct actions.
• Experience translating ambiguous customer requirements into technical scope and delivering against agreed acceptance criteria.
• Evidence of independent technical judgment, effective code review, clear communication, and knowledge transfer.

Preferred experience

• WhatsApp Business Platform, Meta Cloud API, or Twilio integrations, including templates, delivery tracking, consent, and escalation.
• Commerce, payment, booking, or other transactional systems with strict correctness requirements.
• Cloud Run, Cloud SQL, Pub/Sub, and Vertex AI.
• Model training, fine-tuning, or inference optimization using PyTorch, TensorFlow, Hugging Face, or related tooling.
• Previous forward-deployed, founding-engineer, or customer-facing implementation experience.
• Delivery of shared product capabilities derived from multiple customer deployments.

Performance expectations

Success will be assessed against agreed technical and business outcomes:

• Reliable completion of supported customer workflows, with backend-confirmed actions and effective human escalation.
• Measurable improvement in agent quality, failure rates, latency, and operating cost.
• Predictable delivery of maintainable capabilities that can be reused across merchants.
• Clear production visibility, effective incident diagnosis, and tested recovery procedures.
• Stronger engineering practices and sufficient knowledge transfer for another engineer to maintain and operate the system.

Specific targets will be established from the platform baseline and the priorities agreed for the role.