Artificial Intelligence Engineer

Photon; Belgique; Sur place; EUR 90 000 - 140 000; Plein temps

Résumé du poste

Photon seeks an experienced software engineering lead to own end-to-end delivery of LLM-powered applications, including SDK patterns, API integrations, RAG architectures, and guardrails in a regulated financial-services context.

You will set standards for prompt engineering, manage a remote offshore team, and ensure latency, observability, and cost discipline while shipping reliable AI features on schedule.

Qualifications

• 6+ years in software engineering with 2+ years leading end-to-end LLM/AI delivery.
• Proven track record shipping AI features on committed timelines with hands-on troubleshooting.
• Strong Python skills at architectural level.
• Experience architecting LLM API integrations and SDK abstractions across providers.
• Judgment on model selection considering cost, latency and capability trade-offs.
• Deep expertise in prompt engineering and context window management at scale.
• Experience with RAG systems, vector databases, and knowledgebase design.
• Experience designing AI agents/multi-agent patterns and tool-use architectures.
• Strong understanding of guardrails, safety, PII, and compliance in AI outputs.
• Experience defining evaluation frameworks and integrating AI testing into CI/CD.
• Proven ability to design for latency, observability, and cost in production AI systems.
• Financial-services or regulated-industry exposure preferred.
• Strong stakeholder communication and offshore team management skills.
• Bachelor's degree required.

Responsabilités

• Own delivery of LLM API integration and SDK patterns used across applications.
• Define standards for advanced prompt engineering and context window management.
• Own delivery of RAG systems, including vector database topology and knowledgebase design.
• Drive delivery of AI agent and multi-agent patterns and tool-use integration.
• Own guardrails delivery for safety, compliance, and PII handling in prompts/outputs.
• Define evaluation frameworks and real-time eval strategy; set CI/CD AI testing standards.
• Own latency profiling, observability, and cost tracking for LLM-backed systems.
• Run day-to-day offshore development team: standups, blocking, code/design reviews, and hands-on work on critical features.
• Report delivery status, risks, and blockers to engineering leadership.

Connaissances

LLM API integration

SDK architecture

Prompt engineering

Context window management

RAG systems design

Vector database design

AI latency optimization

Offshore team leadership

Formation

Bachelor's Degree

Outils

LangChain

LlamaIndex

Semantic Kernel

AWS Bedrock

MCP (Model Context Protocol)

LLM Gateways

Description du poste

Owns end-to-end delivery of how the organization builds LLM-powered applications: SDK/integration architecture, retrieval and agent design, guardrails, and evaluation. Makes the calls on which LLMs to use for which use cases and sets the observability/cost discipline around AI systems, but is measured on shipped outcomes — runs the offshore team day-to-day and stays hands‑on to unblock delivery risk.

Key Responsibilities

• Own delivery of LLM API integration and SDK patterns used across applications.
• Set organizational guidance on which LLM to use for what use case and drive delivery of multi-LLM scenarios.
• Define standards for advanced prompt engineering and context window management.
• Own delivery of RAG systems, including vector database selection/topology and knowledgebase design.
• Drive delivery of AI agent and multi-agent systems and tool-use/MCP integration patterns.
• Own guardrails delivery (safety, compliance, PII handling in prompts/outputs) — critical in a financial-services context.
• Define evaluation frameworks and real-time eval strategy; set standards for AI testing in CI/CD.
• Own latency profiling, AI observability, and cost tracking/management for LLM-backed systems.
• Run day-to-day delivery of the offshore development team: sprint commitments, code/design review, real-time unblocking, and hands‑on work on critical-path AI features.
• Report delivery status, risks, and blockers to engineering leadership.

Must-Have Qualifications

• 6+ years in software engineering, with 2+ years as a tech lead owning end-to-end delivery of LLM/AI-powered systems (not a pure design/review architect role).
• Proven track record of shipping AI-powered features on committed timelines, including hands‑on troubleshooting under delivery pressure.
• Strong, hands‑on Python skills at an architectural/systems level.
• Proven experience architecting LLM API integrations and SDK-level abstractions across multiple providers.
• Demonstrated judgment on model selection (cost, latency, capability trade-offs) across use cases.
• Deep expertise in prompt engineering and context window management at scale.
• Proven design experience with RAG systems, including vector database architecture and knowledgebase design.
• Experience architecting AI agents/multi-agent systems and tool-use patterns (MCP or equivalent).
• Strong understanding of guardrails design — content safety, PII protection, compliance controls for AI outputs.
• Experience defining evaluation frameworks and integrating AI testing into CI/CD.
• Proven ability to design for latency, observability, and cost management of AI systems in production.
• Financial-services or regulated-industry experience strongly preferred given compliance/guardrail stakes.
• Strong stakeholder communication; able to directly manage day-to-day delivery of an offshore team (standups, unblocking, sprint accountability).
• Minimum Educational Qualification - Bachelors Degree

Nice-to-Have Qualifications

• Direct experience with specific frameworks (LangChain, LlamaIndex, Semantic Kernel, or equivalent).
• Experience with AWS Bedrock or comparable managed LLM platforms.
• Contributions to or deep familiarity with MCP (Model Context Protocol) implementations.
• Experience building internal LLM gateways.
• Familiarity with responsible‑AI/model-risk-management frameworks used in financial services.