Software Engineer II — Agentic AI Foundations

The Agentic AI Foundations team is building the core platform, systems, and primitives that enable Socure to transition from traditional software workflows to agent-native operations. As a Software Engineer II on the team, you will help design, build, and harden a secure, evaluable, vendor-agnostic agent platform that teams across Socure can build on, working alongside senior and staff engineers who set the architectural direction. This is a hands-on, zero-to-one team, and you’ll get outsized exposure to how agentic systems are architected and operated in production.

Responsibilities

Build components of a vendor-agnostic agent platform — including orchestration, tool use, memory, and runtime systems — under the guidance of senior engineers on the team; Implement evaluation and reliability tooling, including metrics, harnesses, and pipelines, to measure and improve agent performance, robustness, and safety in production; Help implement safety and governance controls, including guardrails, policy enforcement, and human-in-the-loop review mechanisms; Build data grounding, retrieval, and memory components that keep agents accurate, context-aware, and aligned with Socure’s domain knowledge and policies; Prototype and iterate on agent behaviors, including planning, multi-step execution, and coordination of tools and services, using real internal workflows as proving grounds; Partner with product and engineering teams to implement agent-powered workflows using the platform primitives the team builds; Apply and help refine documented best practices and design patterns for secure, observable, and scalable agent systems; Bring strong foundational knowledge of LLMs, GPU computing, and model serving to technical discussions and implementation decisions

Requirements

Bachelor’s or Master’s degree in Computer Science, Computer Engineering, Machine Learning/AI, or a related field from top tier institutions, or equivalent practical experience demonstrating strong foundations in computer science and machine learning; 2+ years of professional software engineering experience, with demonstrated experience in distributed systems, backend platforms, infrastructure, or comparable technical environments; Very strong foundational knowledge of large language models and agentic AI systems, including architectures, prompting and orchestration patterns, tool use, and evaluation approaches; Strong foundational understanding of GPU computing and model-serving infrastructure, such as CUDA, vLLM, Ollama, LLMLite, TensorRT-LLM, Triton Inference Server, or similar technologies, including the performance and cost trade-offs associated with serving LLMs at scale; Solid grounding in distributed systems fundamentals, including concurrency, fault tolerance, observability, and performance; Proficiency in at least one modern backend programming language and ecosystem, such as Java, Go, Python, or similar, with comfort working with cloud-native infrastructure, APIs, and data services; Ability to work productively in ambiguous, early-stage problem spaces with guidance from senior engineers, translating direction into working software; A track record of strong technical performance demonstrated through professional impact, research, challenging technical projects, open-source contributions, internships, or other relevant work; Strong collaboration and communication skills, with comfort working alongside cross-functional partners such as product, data science, platform, and security

Nice to have

Experience with multi-agent systems, workflow orchestration, or distributed coordination frameworks through professional work, research, coursework, or technical projects; Experience building or using agent platforms — such as orchestration frameworks, tool registries, or memory systems — or LLM routing, caching, or fine-tuning pipelines through professional work, research, internships, open-source contributions, or personal projects; Exposure to evaluation frameworks, experimentation platforms, or ML systems, such as offline/online evaluations, A/B testing, or agent and model benchmarking; Experience with AI safety, security, or policy systems — including guardrails, policy engines, content filters, or responsible AI frameworks — through professional work, research, coursework, or technical projects; Experience with retrieval systems, knowledge graphs, or data platforms used to ground LLMs and agents in enterprise contexts; Demonstrated depth in ML systems or LLM infrastructure through professional impact, research, publications, technical projects, competition results, open-source contributions, or comparable experience