Application SW Engineer, Vertical AI Agents

Position Overview

Application Software Engineers in this role build vertical AI agent systems that validate and prove the underlying hardware and software capabilities through real-world enterprise deployments. You will operate simultaneously as a product engineer (building compelling agent applications), a platform engineer (stress-testing hardware through realistic workloads), and a security engineer (implementing privacy-preserving inference via FHE hardware acceleration) to influence requirements conversations and go-to-market efforts in regulated sectors such as healthcare and finance.

Key Responsibilities

Design and build enterprise-grade multi-agent AI workflows in target verticals (healthcare, finance, legal) using LangGraph, CrewAI, and/or AutoGen, optimized to leverage advanced CXL memory and accelerator hardware.

Architect agent systems that exploit ultra-long context windows (1M+ tokens) backed by multi-terabyte memory, enabling persistent long-term agent memory and cross-session state beyond typical GPU-based limits.

Implement privacy-preserving inference pipelines using FHE hardware acceleration so agents can process sensitive data (e.g., medical or financial) without exposing plaintext to the inference engine.

Build the agent workload benchmark suite used to validate hardware performance metrics (TTFT, throughput, KV-cache utilization) in coordination with the system software team.

Develop agent security hardening features: least-privilege tool access, skill signing and attestation, behavioral anomaly detection, and prompt-injection defenses aligned with OWASP Top-10 LLM risks.

Prototype multi-agent orchestration for reference deployments in priority regions, validating real multi-agent workloads on the platform.

Collaborate with the system software team to surface agent-level requirements for KV-cache and OKC-style APIs, closing the feedback loop between application behavior and hardware optimization.

Create technical documentation, reference architectures, and integration guides for enterprise and hyperscaler partners.

Required Skills & Experience

6+ years in software engineering, including 2+ years building production multi-agent or agentic AI systems.

Hands-on proficiency with at least two of: LangGraph, CrewAI, AutoGen, LlamaIndex Workflows, or comparable multi-agent orchestration frameworks.

Demonstrated experience validating or benchmarking AI hardware through real-world agent workloads rather than only synthetic benchmarks.

Strong Python engineering skills and experience deploying LLM inference services in containerized / cloud-native environments (e.g., vLLM, SGLang, Triton).

Deep understanding of 1M+ token context optimization challenges: memory management, chunked processing, hierarchical summarization, and large-scale RAG.

Working knowledge of FHE concepts and privacy-preserving ML, with the ability to integrate hardware-accelerated FHE libraries (such as OpenFHE or Concrete-ML) into inference pipelines.

Familiarity with the AI agent security threat landscape: prompt injection, tool misuse, credential theft, and multi-turn escalation-style attacks.

Preferred Qualifications

Domain experience in healthcare AI (e.g., HIPAA/HITRUST), financial services AI (e.g., SOX, GLBA), or other regulated enterprise AI environments.

Experience integrating AI agents with enterprise data systems (EHR/EMR, CRM, ERP) and defining production-ready tool schemas.

Background in AI red-teaming, adversarial prompting, or LLM security research.

Familiarity with regional enterprise or healthcare market dynamics relevant to key go-to-market partnerships.