Applied AI Engineer

Cyberjaya, Malaysia | Posted on 23 / 09 / 2025

Overview

Deriv is an AI-first company, transforming trading through cutting-edge technology. We are seeking a highly skilled Applied AI Engineer to join our dynamic team. You will play a key role in designing, developing, and deploying AI-driven solutions that power our internal tools, marketing automation, and security infrastructure.

Responsibilities

• Design, build, and operate agentic LLM applications with tool-use / function calling, planning, memory, and structured outputs with versioned prompts.
• Implement RAG pipelines (hybrid dense+sparse retrieval, reranking); manage vector indices and the embedding lifecycle.
• Integrate models with internal services and third-party APIs via tool adapters; enforce schema contracts and access controls.
• Lead feature work in recommendation / ranking, content generation / editing, and workflow orchestration backed by measurable eval metrics.
• Build AI-driven security automations : classifiers / anomaly detectors, red-teaming harnesses, and (where appropriate) autonomous testing agents.
• Implement safety & privacy controls : prompt-injection / jailbreak defenses, PII redaction, policy filtering, and auditability.
• Stand up real-time tracing & alerts on token / latency / cost, retrieval degradation, drift, and hallucination risk.
• Own model / prompt / agent versioning and ship via CI / CD with canaries / A-B, rollback plans, and clear SLOs (quality, latency, cost).
• Evolve workflows : PR-based changes to prompts / configs, experiment tracking, concise design docs, and ADRs (decision records).
• Enforce evaluation-first development : golden sets, synthetic probes, regression tests for prompts, grounding / faithfulness checks.
• Review for determinism, observability, and reproducibility; address tech debt with a bias to stability and measured impact.
• Partner with platform / data teams on data contracts, retrieval standards, and tenancy / access control.
• Mentor peers on agent patterns, RAG, prompt engineering, and safe rollout; lead design reviews and share practical research findings.
• Benchmark and integrate new model families / tools via standardized bake-offs; report quality–latency–cost trade-offs.
• Optimize serving with adapters / LoRA, distillation / quantization, routing / caching / speculative decoding; right-size infra for cost / perf.
• Track industry and research trends; convert into pragmatic, incremental stack upgrades.

Requirements

Nice to Have

Benefits & Culture

We pay competitive salaries based on your skills, qualifications, experience and market rates. Additionally, you’ll be eligible for an annual bonus based on individual and company performance.

When you join Deriv, you will be supported in growing your career in our company. You can expect to have a personalised Learning and Development programme as well as the costs covered for relevant professional development and education related to advancing your career at Deriv.

You’ll be part of our Cyberjaya office, home to 600 talented professionals driving innovation in online trading. Our purpose-built, five-storey building enhances collaboration and creativity, leveraging cutting-edge AI and no-code / low-code platforms to push boundaries. You’ll join a vibrant workplace where weekly knowledge-sharing sessions, cross-functional collaborations, and cultural celebrations create an engaging atmosphere. You’ll work in a stimulating, high-tech environment featuring modern conference rooms, a fully-furnished gym, and barista-made coffee, all designed to keep you motivated and inspired.

Our culture is unique, and we live by our values and leadership principles, so it’s worth learning more about Deriv’s culture.

Deriv is an equal opportunity employer committed to diversity and inclusion. We take pride in our Great Place to Work and IIP Platinum certifications, which reflect our commitment to creating an exceptional work environment. Join us and help build the future of online trading while working alongside top talent from over 96 nationalities.

#J-18808-Ljbffr