AI/ML Engineer

Xuman.Ai is live on both app stores. Real people are using it. What is Xuman.Ai Xuman is building a network for on-demand human connection. Billions of people have knowledge, skills, and experience that could be valuable to someone else. Making that accessible is a coordination problem. What you want to do, who’s up for it, and when you’re both available. Xuman brings those things into one system. You discover the person, book their time, and connect. If they’re available now, that can happen immediately. You can also make your own capabilities available through the same network. The goal is to make finding and connecting with the right human as easy as finding information. Think about how much more people could learn, build, and accomplish if that worked globally. The stage What's the rarest stage to join a startup? Building from zero? Everywhere. Joining after scale? Everywhere too. The stage in between is where it gets interesting. The product is built. It's live. People are using it. Now it's about syncing what they feel with what we build. What you'll own The agentic layer in production: agents, tool calling, orchestration, memory, guardrails, on both stores. The architecture and the eval standard, so every change is proven and the team ships with confidence. Model strategy: which models, where, at what cost and latency. Fine-tuning when it earns its place. The data and ML foundation: the signals captured, the pipelines, the first learned models when the data allows. The AI engineers on the team, their sequence and the bar they hold. How we work The founder sets what the product does and how it should feel. When you spot a gap, a bug or a better way, you bring it up before you build it. Once it's agreed, the technical approach is all yours. That is how we stay fast without breaking what people already use. What you bring Engineering experience with real AI or machine learning work behind you, shipped to production. You have shipped agentic, LLM-powered products to real users, not demos. You have trained or fine-tuned a model and put it in production, and you can say what the numbers meant. Multi-agent orchestration you designed in production, including failure paths and recovery. Evals you built yourself, so a change is proven rather than asserted. Python in production, and comfort in a TypeScript service. Ours is a NestJS gateway. Real-time or voice pipelines, or the judgement to build one inside a latency budget. Safety and moderation you have operated, with the false positives that come with it. Model cost treated as an engineering problem you have solved before. You follow the field as it moves, and you have changed an architecture because of what shipped last quarter. You have led a small team of engineers and you still write the hard part yourself. You already build with Claude, Copilot or ChatGPT every day. What helps further data pipelines at real volume. recommendation or matching systems in production. behavioural or cognitive-science exposure. What you get Work that is yours end to end, not tickets off someone else's board. Your work in real hands the same week, on a product that is live today. Pay that grows with every milestone we hit. Growth that is claimed, not handed out. More opens as you carry more. You're joining before the growth. What you build now is the foundation, and inside the company it will always be known as yours. How to be a Xuman Experience the product first. It costs you a few minutes, nothing else. Then pick your path. #J-18808-Ljbffr XUMAN.AI