Job
Description
🚀 What we’re building CodeRound AI matches top 5% tech talent to fastest growing VC funded AI startups in Silicon Valley and India. Candidates apply once and get UPTO 20 remote as well as onsite interview opportunities IF selected! T op-tier product startups in US, UAE & India have hired top engineers & ML folk using CodeRound 🧩 What you'll do Design, build, and iterate on LLM-based systems that score freeform answers, generate real-time feedback, and assist. Optimize prompt workflows and build tool-augmented agents that evaluate coding, reasoning, and system design responses. Implement and tune RAG pipelines to personalize interviews, fetch dynamic context, and reduce hallucinations. Define LLM evaluation frameworks — accuracy, consistency, bias, and interpretability. Collaborate with backend and product to ship fast, test with users, and refine in production. Keep up with SOTA papers, models, and tools — and bring them to life in real applications. ✅ You’ll thrive here if you: Have built LLM-based applications in production — not just toy projects. Are strong in Python, familiar with LangChain/LlamaIndex, and have worked with OpenAI, Claude, or open-weight models. Understand LLM failure modes — and how to prompt, chain, or retrieve around them. Think like a product engineer — fast, iterative, user-focused, and not afraid to ship experiments. Are excited about building LLM features that go beyond chat — evaluation, scoring, ranking, summarization. ⚡ Bonus if you: Have experience with vector DBs (Pinecone, Weaviate, etc.), semantic search, and evaluation frameworks. Have worked on AI in edtech, hiring, developer tools, or assessment. Have finetuned models or built systems around LLM agents + tools. Want to grow into a Founding AI Lead role. ✨ Why CodeRound? You’ll own mission-critical LLM systems used by top startups You’ll work with a fast team of engineers and founders who’ve built and scaled startups before. You’ll ship fast, see real user feedback, and influence product + infra decisions from Day 1. Show more Show less
