AI Engineer – Hermes Agent SME in Bangladesh

Key Responsibilities

• Enterprise System Orchestration: Integrate the Hermes Agent framework with modern ERP systems and global trade management solutions to automate complex back-office, financial, and logistics workflows.
• Regulatory & Authority Integration: Architect secure, sovereign interfaces for government and authority portals, ensuring automated compliance reporting and data exchange that meets strict regulatory standards.
• Hermes Agent Management: Design and manage isolated Hermes Profiles, including the configuration of SOUL.md, persistent memories, and custom skill-sets tailored to specific departmental needs.
• Skill Synthesis: Develop and refine autonomous skills using the agentskills.io standard, enabling agents to synthesize experience from complex enterprise tasks into reusable tools.
• MCP Standardization: Leverage the Model Context Protocol (MCP) as a universal standard to connect autonomous agents with external databases, APIs, and proprietary enterprise applications.
• Self-Evolution via GEPA: Implement the GEPA Self-Evolution system to optimize internal procedures and prompts based on execution logs, reducing operational overhead and token consumption.

Required and Preferred Qualifications

• Hermes Agent Mastery: Proven experience in configuring and scaling the Hermes Agent framework (including memory backends like HindSight) for persistent, session-spanning autonomy.
• Enterprise Integration Expert: Deep technical understanding of connecting AI agents to ERP infrastructures and global trade systems via secure APIs and microservices.
• Sovereign Infrastructure: Strong commitment to local-first AI and digital sovereignty, with hands-on experience deploying models via vLLM or Ollama on self-hosted GPU clusters.
• Compliance Knowledge: Familiarity with the technical requirements of the EU AI Act and GDPR, specifically regarding transparency, auditability, and human-in-the-loop mechanisms in automated decision-making.
• Python Proficiency: Advanced software engineering skills in Python, utilizing async patterns and typed schemas (e.g. Pydantic) to build resilient agentic cycles.

Nice to Have

1. Experience with LangGraph, dify, or Dagster for orchestrating multi-step, cross-departmental workflows.
2. Familiarity with vector databases (e.g. Qdrant, PGVector) for high-performance retrieval-augmented generation (RAG) in corporate knowledge bases.
3. Background in Infrastructure as Code for managing sovereign cloud deployments.