AI Full-Stack Engineer Intern
Company: Vela (YC W26)
Location: Location not specified (Remote)
Remote: Yes
Posted: 2026-02-02
About this role
Position Details
- Pedigree: Insanely cracked and gets shit done
- Education: Bachelors in Computer Science
- Employment Type: Full-time position
- Work Arrangement: Remote
Core Responsibilities
AI & Agent Systems
- Design, build, and iterate on autonomous AI agents that handle complex, multi-step workflows and manage complex context states
- Implement robust reasoning systems that handle ambiguity, edge cases, and real-world messiness
- Build evaluation frameworks to measure agent reliability, reasoning quality, and failure modes
Full-Stack Development
- Own the frontend experience (React, TypeScript) ensuring AI capabilities translate into intuitive UX
- Design and implement backend services and APIs for real-time AI processing
- Build data pipelines that feed context to AI systems while maintaining privacy and security
Infrastructure & Scale
- Take ownership of system reliability, observability, and performance as we scale
- Implement security best practices for handling sensitive user data and third-party integrations
- Design for growth: architect systems that work at 1K users and evolve gracefully to 1M+
Technical Requirements Essential Technical Skills
- Deep understanding of GenAI/AI concepts including:
- Large Language Models & Prompting: How LLMs work under the hood (attention, tokenization, context windows), prompt engineering patterns, and their failure modes. You should be able to debug why an LLM is producing bad outputs and fix it.
- Embeddings & Vector Search: Practical experience with text embeddings (OpenAI, Cohere, or open-source models), vector databases, and semantic retrieval. Understanding of embedding space geometry, similarity metrics, and when vector search fails.
- Agent Architectures: Familiarity with patterns like ReAct, chain-of-thought, tool use, and multi-agent coordination. Experience with at least one agent framework (LangChain, LangGraph, AutoGen, CrewAI, or similar).
- Evaluation & Reliability: Ho...