Agentic AI Intern

Role Overview

SoftworkerAI is an early-stage AI startup building a work platform that helps individuals and teams delegate, manage, and execute work with AI-powered agents. We are looking for an Agentic AI Intern to research, design, evaluate, and prototype advanced AI agent systems for planning, reasoning, memory, orchestration, tool use, workflow automation, and human-AI collaboration. This role is ideal for someone who is deeply curious about how AI agents can move beyond chat interfaces and become reliable systems.

Responsibilities

• Research agentic AI architectures, including planning, reasoning loops, tool use, memory, context engineering, RAG, orchestration, and multi-agent systems.
• Study and evaluate open-source agent frameworks, SDKs, memory systems, vector databases, graph databases, workflow engines, and AI-native infrastructure.
• Prototype agent workflows for task execution, human-in-the-loop approvals, long-running jobs, retries, recovery, and state management.
• Design and test memory systems including short-term memory, long-term memory, episodic memory, semantic memory, workspace memory, and user-specific context.
• Explore hybrid RAG, GraphRAG, knowledge graphs, retrieval pipelines, reranking, metadata filtering, and context compression.
• Evaluate agent reliability, hallucination risks, task completion quality, latency, cost, and failure modes.
• Create research notes, technical briefs, system design documents, architecture diagrams, and implementation recommendations.
• Collaborate with product, engineering, and design teams to translate research insights into practical product features.
• Track developments in agentic AI, LLM infrastructure, AI developer tools, and emerging research papers.

Requirements

• Strong understanding of LLMs, AI agents, RAG, embeddings, tool calling, and prompt/context engineering.
• Ability to read research papers, technical documentation, GitHub repositories, and system design material.
• Interest in memory systems, orchestration frameworks, multi-agent workflows, AI-native operating systems, and human-AI collaboration.
• Ability to think from first principles and compare trade-offs across different architectures.
• Comfortable writing clear research summaries, technical documents, and prototype specifications.
• Basic programming ability in Python, TypeScript, or Rust is preferred.
• Strong ownership, curiosity, and ability to work independently in a fast-moving startup environment.

Skills

• Python
• TypeScript
• LangChain
• RAG
• Vector Databases