AI Product Engineer (Backend and LLM)

AI Product Engineer (Backend × LLM) We are building the "Learning OS" - the AI intelligence layer for education, with the same engine expanding into how organizations understand their own knowledge and people. We're looking for a hungry, product-minded engineer who lives at the intersection of solid backend engineering and applied LLM work. You'll work on genuinely hard problems: deep understanding of multimedia content, high-precision OCR, and orchestrating reliable multi-stage pipelines that deliver 100x cost savings vs. direct API calls to big frontier models. And you'll ship them as real products - services, APIs, queues, storage, observability - not notebooks. Who You AreA backend engineer who went deep on LLMs - or an LLM engineer who can actually ship a backend. Our projects are products. They need real services behind them, but every core feature is an LLM feature. You should be comfortable owning both sides: designing an async pipeline that doesn't fall over at scale and the prompt/eval loop that makes its output trustworthy. Builder. High agency, owns problems end-to-end, reliably ships working production systems. LLM-aware, not LLM-naive. You know when a well-engineered prompt beats a custom model - and where LLMs fail. The best solutions mix classical approaches (deterministic parsers, retrieval, heuristics, graphs) with LLMs to cover their gaps. Strong learner. You don't need to be an expert, but you should already be reading papers and following open-source model releases (DeepSeek, Qwen, …) out of genuine interest. Data quality obsessed. You'll manually label 100 samples when needed - and you also care about proper evaluation (we have resources to provision at-scale human evals). RequirementsMust-haves Strong Python and real backend experience: designing and shipping production services (APIs, async/queued workloads, databases, caching, deployment, monitoring).Hands-on with LLM APIs and prompt engineering in production.Database design and optimization - schemas, indexing, query tuning, and performance debugging (SQL and Vector/NoSQL).Data engineering basics: PDF/image extraction, and the nuances of working with JSON and markup languages.Comfort with messy, unstructured, real-world data and the schemas needed to tame it. Nice-to-haves Cloud experience with GCP or Azure - deploying and operating production services, managed databases, container orchestration, and GPU compute.Evaluation methods and eval infrastructure.Fine-tuning experience.Agentic pipelines, tool use, multi-step orchestration.Retrieval systems, knowledge graphs, or structured representations of entities and their relationships.Enterprise-grade B2B systems for international clients - multi-tenancy, permissions, integrations, data privacy. More important than any checkbox: extreme learning velocity, bias toward action, rigorous experimentation, and genuine curiosity about why things break. What You GetCompetitive pay, accelerated growth working directly with experienced engineers, full end-to-end ownership, and high-impact work on bleeding-edge multimodal AI.