Principal Forward Deployed Engineer - AI PlatformPrincipal Forward Deployed Engineer - AI Platform/Kubernetes/Pytorch

Job description

About the Job: As a Principal Forward Deployed Engineer (FDE) within the APAC AI Center of Excellence, you will operate at the absolute technical frontier of Enterprise and Sovereign AI. This is a high-impact, engineering-led adoption role for a recognized technical expert who drives innovation across teams and connects our core platform capabilities with opportunities to create significant customer value. You will write production-grade code directly alongside our most strategic customers, developing custom platform extensions, defining overarching technical strategies to solve business-critical problems, and architecting future-proof systems that set new standards for reliability, security, and performance. Following our open-source DNA, your mission is two-fold: solve the immediate, high-stakes technical problems of APAC enterprises and governments, and ensure that 70% of the primitives and features you build in the field are successfully upstreamed into global open-source communities (e.g., distributed runtimes, orchestration engines, and MLOps platforms) and integrated back into core enterprise AI products. Note: This role may come into contact with confidential or sensitive customer information requiring special treatment in accordance with company policies and applicable privacy laws. What you will do? 1. High-Bandwidth Engineering, Systems Architecture & QualityOvercome the "Integration Wall": Embed with strategic customers and partners to resolve deep-seated engineering blockers (such as custom identity access management/SSO, legacy data pipelines, and network isolation/air-gapped challenges). Architect Future-Proof Solutions: Drive the technical strategy and design of software solutions across multiple subsystems, influencing the overall architecture of customer AI platforms. Ensure Software Quality & Reliability: Establish, maintain, and monitor testing practices for large-scale AI systems. Ensure all field-developed integrations follow robust architectural patterns to deliver superior software quality and long-term operational stability. Rapid Prototyping & Integration: Rapidly prototype and develop custom, secure platform integrations using customers’ real-world data, validating complex distributed inference systems, advanced Retrieval-Augmented Generation (RAG) pipelines, and autonomous AI agent architectures. Last-Mile Engineering: Harden initial Proofs-of-Concept (POCs) and Minimum Viable Products (MVPs) into secure, robust, and highly scalable production architectures. 2. Global R&D Bridge, Product Feed Loop & CollaborationGlobal Engineering Collaboration: Act as the critical technical bridge between APAC market requirements and global engineering teams. Seamlessly collaborate across product and ecosystem engineering hubs globally to harden field innovations into core enterprise AI products, align on partner blueprints (such as hardware-optimized models, containerized inference runtimes, and distributed scheduler workloads), and drive collaborative developments on cutting-edge agentic frameworks. Community Leadership: Drive innovation by leading significant product-area initiatives with a community-first mindset. Participate across multiple upstream communities, foster and monitor community health, and engage in industry and internal working groups. Telemetry and Performance Tuning: Feed back real-world performance telemetry, operational edge cases, and security evaluation data (using automated LLM vulnerability assessment and red-teaming tools, e.g., Garak, Chatterbox) to optimize global core product guardrails and engine designs. 3. Open Source Upstream Stewardship & SDLC EvolutionDirect Contribution: Commit upstream code directly to core open-source projects under the cloud-native AI, model serving, and MLOps umbrella. Technical Debt Elimination: Refuse to build bespoke, isolated customer code. Standardize regional solutions so they are structurally viable for upstream merging, preventing regional code divergence and downstream technical debt. Evolve Development Practices: Drive the evolution of the Software Development Life Cycle (SDLC) within the organization, introducing new methodologies, tools, and best practices that improve the efficiency of multiple teams contributing to large-scale systems. 4. Leadership, Mentorship & Business ImpactMentor and Develop Engineering Talent: Coach and mentor junior and principal-level engineers across the organization. Role-model technical mentorship, active listening, and open-source stewardship to raise the technical bar across APAC. Own and Deliver Business Impact: Own and drive key technical initiatives, recognizing how distinct platform components flow together to deliver maximum subscription and operational value to the end user. Drive Strategic AI Automation: Formulate the strategy and best practices for integrating advanced AI ecosystems to automate and optimize large-scale operational systems. Focus Areas & Specialization TracksCandidates will be expected to demonstrate deep capability in at least one of the following two tracks: Track A: Hardware Heterogeneity & Sovereign AIFocus on kernel-level and runtime-level tuning for domestic and specialized NPUs/GPUs (e.g., emerging East Asian hardware accelerators). Develop hardware-optimized inference engine plugins (such as vLLM-compatible backends) and drivers to enable a highly diversified hardware ecosystem. Build highly localized, sovereign, on-premises AI deployment stacks ensuring strict regulatory, residency, and security compliance. Track B: AI Platform Primitives & Distributed SystemsOptimize hyper-scale distributed inference systems using advanced scheduling techniques (e.g., disaggregated prefill/decode, KV cache offloading, and dynamic autoscaling). Maintain and develop localized enterprise data pipelines, contributing complex parsing workflows back to open-source document parsing frameworks (such as Docling) and synthetic data generation tools. Implement secure Agentic AI lifecycles (AgentOps), testing and refining cloud-native agentic operators (such as Kagenti) alongside strict workload identity frameworks and API gateway protocols (e.g., Model Context Protocol) What you will bring? Technical RequirementsSoftware Engineering: Exceptional proficiency in C/C++, Go, and Python. Must have a proven track record of shipping production-ready, highly optimized, and robustly tested code. Upstream Contribution: Demonstrated status as an active upstream contributor or maintainer in key open-source communities (such as deep learning engines, distributed computing schedulers, or container orchestration runtimes). AI/ML Infrastructure: Hands-on experience with deep learning frameworks, model fine-tuning (LoRA, QLoRA, SFT), large-scale model serving architectures, and LLM orchestration (e.g., LangChain, LlamaIndex). Cloud-Native Mastery & Architecture: Strong experience architecting on Kubernetes or enterprise container platforms, including writing and managing Custom Resource Definitions (CRDs), custom operators, and multi-subsystem integrations. Hardware & Acceleration: Familiarity with hardware-level optimizations, CUDA, ROCm, driver compilation, and GPU/NPU operator configuration. Professional Experience & Soft SkillsExperience: Minimum of 8+ years of experience in system engineering, distributed computing, platform engineering, or AI/ML software development, with a clear history of technical strategy leadership. Technical Mentorship: Demonstrated experience mentoring senior technical staff and driving modern software delivery practices (SDLC, CI/CD, and robust testing frameworks). Ambiguity Tolerance: High capacity to navigate ambiguous, rapid-velocity environments within a large enterprise structure. Enterprise Presence & Policy Mindfulness: Excellent communication and consultative skills; ability to engage with enterprise customer architects and C-suite technologists while remaining highly mindfu