Website:
amunra.io
Job details:
Position Overview
We are seeking a Senior AI Platform & Systems Engineer to build the operational backbone connecting our proprietary complexity science models with client-facing autonomous AI agents.
This role bridges dynamic simulation modeling, modern agent protocols, and distributed enterprise platforms across three core areas:
1. Context & Memory Architecture: Architecting a persistent, stateful knowledge plane (Knowledge Graphs + Hybrid Vector Retrieval) that structures the causal relationships, historical simulation runs, and state-space dynamics of our models.
2. Model Context Protocol (MCP) Integration: Exposing our complexity models and context substrate as standardized, low-latency MCP servers and deterministic agent tool suites.
3. Private Enterprise Delivery: Packaging and delivering this infrastructure to institutional clients via isolated environments, private networking (e.g., AWS PrivateLink), and hardened multi-tenant security layers.
Key Responsibilities
- Context & Memory Architecture: Design and deploy a stateful knowledge substrate using graph databases (e.g., Neo4j, Memgraph) and vector stores to map complex simulation states, causal loops, and scenario histories.
- Persistent Agent Memory: Engineer long-term episodic and semantic memory architectures that enable autonomous agents to maintain situational awareness and cross-session context without degradation.
- MCP & Tool Interface Engineering: Build production-grade Model Context Protocol (MCP) servers and tool execution interfaces, allowing AI agents to query the model knowledge base, trigger simulation runs, and inspect results deterministically.
- Enterprise Delivery & Isolation: Architect private distribution channels for enterprise clients (AWS PrivateLink, VPC peering, dedicated tenant gateways, mTLS) ensuring zero cross-tenant data or context leakage.
- Client SDKs & Integration Tooling: Develop lightweight, type-safe Python and TypeScript SDKs and API schemas that enterprise engineering teams can integrate into their existing stacks.
- Telemetry & Execution Auditing: Build distributed tracing (OpenTelemetry) and immutable audit logging capturing every agent prompt, retrieved memory node, parameter configuration, and simulation output.
Required Qualifications
- Distributed Backend Systems: 4+ years of production experience building high-throughput, low-latency distributed backends in Python, Go, or Rust.
- Model Context Protocol (MCP) & Agent Tooling: Practical experience implementing MCP servers/clients, structured function calling, or deterministic agent execution runtimes (e.g., LangGraph, Temporal, or custom state machines).
- Graph & Hybrid Retrieval Systems: Hands-on experience with Graph Databases (Neo4j, Memgraph, AWS Neptune) and GraphRAG / Hybrid Retrieval (combining structured knowledge graphs with vector embeddings).
- Enterprise Cloud Security & Networking: Demonstrated experience designing private enterprise connectivity (AWS PrivateLink, Azure Private Link, VPC peering) and zero-trust authentication (mTLS, OAuth2/OIDC, granular RBAC).
- Protocol & Streaming Design: Deep expertise with protocol-level design using gRPC / Protocol Buffers, WebSockets, and asynchronous APIs.
Preferred Qualifications
- Prior experience collaborating with quantitative researchers, computational scientists, or financial engineers to productionise mathematical/simulation models.
- Experience packaging backend runtimes into Helm charts or Kubernetes operators for client-managed VPC deployments.
- Familiarity with high-performance serialisation formats (e.g., Apache Arrow, FlatBuffers).
Click on Apply to know more.
