Agent Solution Architect (Logistics)

Overview

The Prompt Engineer / Agent Solution Architect builds, expands and manages AI agents that power project44's AIA Operations organisation, acting as both an AI workflow architect and operational strategist.

What you'll do

• Design, build and maintain AI agent workflows that automate carrier monitoring, performance tracking, exception management and escalation processes.
• Build custom, customer-specific AI agent workflows tailored to individual carrier, lane or account requirements.
• Write, test, iterate and optimise prompts for large language models, ensuring accuracy, reliability and operational efficiency.
• Run structured tests comparing prompt variants against defined criteria, and document what works, what doesn't, and why.
• Track prompt versions and iterations, maintaining clear records of changes, evaluation outcomes and design reasoning.
• Diagnose and resolve shortfalls in LLM outputs through systematic prompt refinement.
• Work with engineers through handover and deployment, ensuring workflows behave as intended at go-live.
• Ramp up new use cases post-launch, surfacing edge cases and refining prompts as volume increases.
• Monitor agent performance on an ongoing basis, identifying drift, degradation or failure modes and re-iterating on prompts.
• Maintain and support production workflows throughout their lifecycle, applying fixes and updates as customer needs, data sources or models evolve.
• Evaluate new AI tools and platforms and provide structured recommendations on adoption and integration.
• Continuously increase AI agent build velocity by escalating process pain points and driving improvements.
• Identify opportunities to expand automation coverage and bring forward structured proposals for new AI capabilities.
• Read and interpret Product Requirements Documents and transform them into clear AI agent requirements.
• Work with Product Managers across the full agent build lifecycle, flagging gaps or ambiguities early.
• Work with engineers who build the connectors that deploy AI agents, ensuring clear handover of prompt designs and workflow logic.
• Collaborate with external vendors to reduce friction and influence new feature releases.
• Partner directly with customers to understand operational goals and translate them into tailored workflows and success criteria.
• Review agent performance with customers post-launch and drive iterative improvements.
• Serve as subject-matter point of contact for customers on agent behaviour, troubleshooting and enhancement requests.
• Contribute to training resources, internal knowledge bases and team documentation.
• Champion a culture of disciplined experimentation, continuous learning and rigorous quality standards.

What you'll need

• 4+ years in transportation operations, logistics technology, carrier management or network operations.
• Hands-on experience working with AI tools, automation platforms or workflow-based technologies in an operational context.
• Troubleshooting experience across EDI, API, data pipelines, system configuration and workflow orchestration.
• Strong analytical skills, comfortable using data, monitoring dashboards and structured observation to diagnose problems and measure outcomes.
• Demonstrated ability to work collaboratively across Product, Engineering, AI and Operations stakeholders.
• Excellent written and verbal communication skills; able to read product requirements and translate them into clear, structured specifications for AI workflows.
• A genuine curiosity about how LLMs work and a methodical approach to testing and refining prompts.
• Ability to write clear, structured prompts and instructions that direct AI behaviour accurately and consistently.
• Familiarity with core prompting techniques such as step-by-step instructions, few-shot examples and breaking complex tasks into stages.
• Comfort working in AI tools and platforms to build, test and iterate on workflows without needing to write code.
• Ability to critically evaluate LLM outputs, spotting errors, inconsistencies and failure modes, and translate observations into prompt improvements.
• An understanding of LLM behaviour and limitations, including hallucination, misinterpretation of instructions and inconsistent outputs, and how to design prompts that account for them.

Nice to have

• Global or multi-region experience.

Details

• Location: Bengaluru, India.
• Work mode: operate from the office all 5 days a week.
• Shift time: 12:30 PM to 8:30 PM IST.