As a Principal Software Engineer at JPMorganChase within the Chief Data and Analytics Office (CDAO), you provide expertise and engineering excellence as an integral part of an agile team to enhance, build, and deliver trusted market-leading technology products in a secure, stable, and scalable way.
In this role, you will lead the design and evolution of the firm’s GenAI serving platform, focused on high-performance LLM inference, intelligent model routing, and GPU efficiency, to deliver reliable, cost-effective AI capabilities at enterprise scale. Leveraging your advanced technical capabilities and collaborating with colleagues across the organization you will drive best-in-class outcomes across various technologies to support one or more of the firm’s portfolios. Influence leaders and senior stakeholders across business, product, and technology to drive alignment and outcomes; foster a culture of diversity, opportunity, inclusion, and respect.
Job Responsibilities
- Design, build, and operate a high-throughput, low-latency LLM serving platform (batching, scheduling, caching, streaming responses, multi-tenancy, and autoscaling) across GPU/CPU fleets.
- Build and evolve a GenAI Gateway / inference API layer (authentication, authorization, quota/rate limiting, routing, request shaping, policy enforcement hooks, and standardized observability) to support diverse application workloads.
- Develop and optimize “open routing” / intelligent model routing across multiple model backends (open-source and vendor models), balancing quality, latency, reliability, and cost with configurable policies and guardrails.
- Drive GPU serving optimization: kernel-level performance tuning where needed; model compilation/acceleration (e.g., TensorRT-style approaches), efficient memory management, KV-cache strategies, and throughput tuning (prefill vs. decode optimization).
- Implement quantization and compression strategies (e.g., INT8/INT4, weight-only quantization), including evaluation-driven selection and safe rollout practices that preserve quality and reduce cost/latency.
- Design and implement disaggregated serving patterns (e.g., separating prefill/decode, KV-cache offload, tiered serving) and distributed inference architectures to improve utilization and tail latency.
- Develop secure, high-quality production code; review, debug, and improve code written by others; create durable, reusable frameworks and platform components leveraged across teams, aligned to modern product development methodologies.
- Own and support SDK and service integrations, ensuring reliability, performance, and maintainability.
- Establish SLOs/SLAs for inference services and build operational excellence (load testing, capacity planning, incident response playbooks, regression detection, and continuous performance benchmarking); build robust performance and cost observability (latency histograms, token throughput, GPU utilization, memory fragmentation, cache hit rates, per-tenant cost attribution) and automate remediation of recurring issues.
- Architect and govern agentic AI-enabled engineering workflows (using enterprise-authorized tools within the work environment) to improve delivery speed, code quality, and operational outcomes at scale (e.g., AI-driven PR review assistance, test generation/maintenance, release readiness checks, incident triage and root-cause acceleration), while defining guardrails for validation, security, resiliency, and reuse across teams.
- Apply knowledge of tools within the Software Development Life Cycle toolchain, including enterprise-authorized AI-assisted development and automation capabilities, to improve the value realized by automation at scale .
Required qualifications, capabilities, and skills
- Formal training or certification on software engineering concepts with 7+ years applied experience including hands-on delivery of system design, application development, testing, and operational stability for large-scale platforms and services.
- Expert proficiency in one or more programming languages (e.g., Python, Java, Scala, Go) with strong code quality, testing, and debugging practices.
- Demonstrated experience designing and leading adoption of agentic AI-enabled development practices (using enterprise-authorized tools within the work environment) across teams, including setting standards for human-in-the-loop validation, auditability/traceability of changes, and secure handling of sensitive data.
- Strong understanding of responsible AI use and control expectations in engineering workflows, including security/resiliency implications, data sensitivity, and risk-based governance; ability to influence senior technical leaders on safe scaling patterns and reuse.
- Proven experience designing and operating high-scale inference and distributed systems (multi-tenant services, backpressure, load shedding, rate limiting, request prioritization, and tail-latency reduction).
- Strong understanding of GPU-serving fundamentals (compute/memory trade-offs, batching, concurrency, network bottlenecks, and performance profiling) and experience improving efficiency/utilization in production.
- Experience with model serving stacks and patterns (model registries/artifacts, rollout strategies, canaries, A/B, shadow traffic) and performance benchmarking methodologies.
- Practical cloud-native experience (containers, orchestration, IaC, observability) and experience operating production systems with clear SLOs.
- Experience applying new methods to solve complex technology problems across one or more technical disciplines (platform engineering, ML systems, data engineering, distributed systems).
- Strong communication skills: able to present to and influence senior leaders/executives, translating complex technical topics into clear decisions and trade-offs.
- Strong understanding of business outcomes and product delivery, and ability to align platform roadmaps to measurable impact.
Preferred qualifications, capabilities, and skills
- Deep experience with LLM inference optimization techniques (e.g., speculative decoding, KV-cache management, paged attention-style approaches, optimized sampling, continuous batching).
- Practical experience with quantization and compression toolchains (evaluation, calibration, regression testing, production rollout) and understanding of quality/performance trade-offs.
- Experience designing disaggregated serving architectures (prefill/decode separation, cache offload, distributed inference) and operating them at scale.
- Experience building model routing and governance layers (policy-based routing, fallback strategies, circuit breakers, per-tenant controls, cost-aware routing).
- Strong performance engineering background (profiling, flame graphs, GPU profiling, bottleneck analysis) and production tuning under real workload constraints.
- Experience with multi-tenant platforms, reusable frameworks, and developer self-service capabilities at enterprise scale.
- Strong security-by-design experience for ML/LLM systems (secrets, access control, data handling, supply chain controls) and resiliency engineering.