Agentic AI Engineer Intern
TobSar · Phnom Penh, Phnom Penh, Cambodia
Tailor my CV for this job — freeWe’re Hiring: Agentic AI Engineer InternMemory, Tool Use & Autonomous Sales Workflows
We help merchants answer customer questions, recommend products, manage orders, and automate sales workflows across messaging channels such as Facebook Messenger and Telegram.Our goal is to build autonomous sales agents that understand customer needs, remember relevant information, use business tools, and complete tasks reliably—with human support when needed.About the InternshipWe are looking for an Agentic AI Engineer Intern to help build the systems behind reliable, autonomous agents.Working alongside our technical team, you will contribute to agent memory, context management, tool calling, and agent harness engineering—the software that manages an agent’s execution, permissions, progress, and recovery.You will take ownership of scoped features, receive technical guidance and code reviews, and test your work against practical merchant scenarios, including colloquial Khmer and mixed Khmer-English conversations.Key ResponsibilitiesMemory & Context Management: Help agents maintain conversation state, retrieve relevant customer and merchant information, and handle outdated or conflicting information.Tool Calling & Integrations: Connect agents to product catalogs, inventory, order services, and business APIs using validated, structured inputs and outputs.Agent Harness & Reliable Execution: Help implement execution loops, checkpoints, pause-and-resume behavior, retries, execution limits, and duplicate-action prevention.Autonomous Sales Workflows: Develop scoped workflows for product inquiries, recommendations, order intake, and follow-ups.Verification & Human Handoff: Ensure agents verify action outcomes, ask clarifying questions, and request human review when necessary.Evaluation & Debugging: Create test cases, inspect execution traces, and measure task completion, memory accuracy, tool-use correctness, latency, and cost.Backend Development: Contribute APIs, webhook handlers, and background tasks that connect agent workflows to TobSar’s platform.RequirementsCurrently studying or recently graduated in Computer Science, Data Science, Software Engineering, or a related field—or equivalent practical experience.Solid programming fundamentals in Python or TypeScript/JavaScript.Experience building at least one project involving APIs, backend development, automation, or AI integration.Basic understanding of HTTP APIs, JSON, databases, and Git.Familiarity with LLMs and an interest in tool calling, structured outputs, retrieval, and agent memory.Ability to debug problems, explain your approach, and learn from feedback.Ability to communicate in Khmer and read technical documentation in English.Willingness to collaborate onsite with a small startup team.University, personal, and open-source projects are welcome. Previous professional AI experience is not required.Helpful, but Not RequiredExperience with Django, FastAPI, PostgreSQL, Redis, or Docker.Experience with an agent framework or a custom tool-calling loop.Familiarity with asynchronous tasks, webhooks, or event-driven systems.Experience building AI evaluations or Khmer-language applications.Basic React or Next.js skills for merchant-facing tools.What You Will GainHands-on experience building AI agents for real business workflows.Technical mentorship, code reviews, and collaboration with TobSar’s engineering team.Ownership of a scoped project with measurable outcomes.Exposure to memory systems, reliable execution, integrations, and deployment.Potential consideration for a full-time role, depending on performance and company needs.How to ApplyApply through https://hr.tobsar.com/recruitment/open-recruitments for the Agentic AI Engineer Intern position.Please include:Your CV.Your GitHub, portfolio, or a project demonstration.A short explanation of one project you built and your contribution.Your earliest start date, weekly availability, and preferred internship duration.A brief answer to this technical question:“An AI sales agent remembers a customer’s selected product and calls an API to create an order. The API times out, and the agent restarts. How would you help it resume without losing context or creating a duplicate order?”We value clear reasoning. A short explanation or diagram is enough; no code is required to apply.