AI Software Engineer, Agent Harness

About the role

EnCharge AI is building a next‑generation AI platform based on an innovative in‑memory computing architecture. The AI Software Engineer will own the agent harness layer that runs agents against files, tools, and documents, ensuring safe execution, context management, and reliable outputs.

Key responsibilities

• Own the harness architecture end‑to‑end, including agent loop, safe execution, context management, knowledge base, memory, permissions, orchestration, outputs, interfaces, and observability.
• Build custom components that have no open‑source equivalent, such as session semantics, enforced permissions, human‑editable memory files, orchestrator, and output handling.
• Keep pace with evolving models by implementing adapters, prompt formats, tool‑call schemas, stop conditions, benchmarking, and evaluation.
• Ensure reliable tool use across models with varying tool‑calling quality through validation, repair, retries, and fallbacks.
• Develop agents, tools, and MCP servers for internal and customer use cases and perform security reviews before release.
• Create an evaluation harness with task suites, regression runs, and cost/latency tracking alongside quality metrics.
• Define clear interfaces such as session API, CLI, GUI, and endpoint for existing tools.

Required profile

• 10+ years of software engineering experience in backend systems or ML infrastructure.
• Strong Python skills and proficiency in at least one systems language (Go, Rust, or C++).
• Experience shipping and operating an agent loop in production, including multi‑step workflows and unattended runs.
• Hands‑on expertise with Retrieval‑Augmented Generation (RAG), context management, and memory for LLM applications.
• Background in sandboxing, isolation, and permission models for automated systems.
• Practical experience running open‑weight models and understanding how quantization and serving choices affect behavior.
• Comfort working in fast‑moving, ambiguous environments with strong product instincts.

Required skills

• Python
• Go
• Rust
• C++
• Retrieval‑Augmented Generation (RAG)
• Sandboxing
• Isolation
• Permission models
• Quantization
• Model serving