Founding AI Engineer — Agentic Systems

**COMPENSATION — READ THIS FIRST ** Before launch and seed round (approx. months 4–6): equity-only. A founding-team equity grant negotiable based on experience, scope, and contribution. Standard vest and cliff. After launch and seed round: competitive base salary in the $190k–$280k range, ongoing equity participation through refresh grants, and other benefits. Why we lead with this: we value your and our time. We are looking for people who want to make their mark. Astoria is building the most ambitious product in the category. The pre-launch and pre-seed round phase is founder-mode: small team, high intensity, outsized equity. This role is for someone who sees the equity as the primary upside — not a supplement — because they believe what we're building will define the category. **The 90-Second Version ** We are in stealth and will help you with your diligence. Astoria AI is building an agentic runtime for human potential. Multiple specialized AI agents serve candidates and companies across the entire hiring and talent lifecycle: career strategy, networking intelligence, verified matching, compliance-native hiring, onboarding, engagement, retention, and workforce planning. We are attacking a market that is structurally broken in ways that damage real people. 44% of resumes contain fabrications. ATS keyword filters reject 88% of qualified candidates. 27% of posted jobs are ghost jobs. Meanwhile the regulatory wave — Mobley v. Workday, the EU AI Act, FCRA lawsuits against Eightfold — is forcing every enterprise buyer to ask whether their hiring AI creates legal risk or eliminates it. Nobody has fixed this. The incumbents can't, because their business models depend on the broken parts. That is the opening. **What You'll Own ** - Design and build LLM-powered workflows that perform multi-step tasks, not just answer questions. - Build agentic systems that can retrieve context, call tools, use structured outputs, work with data, and escalate to humans when needed. - Build data and intelligence pipelines that collect, normalize, classify, enrich, and score talent, company, role, and market signals. - Work with documents, structured data, product data, APIs, internal knowledge sources, and external signals. - Apply lightweight ML and statistical approaches for classification, ranking, entity matching, deduplication, confidence scoring, and signal quality measurement. - Use RAG and retrieval where helpful, with attention to grounding, source traceability, retrieval quality, and evaluation. - Implement structured prompting, function/tool calling, JSON/schema-based outputs, workflow state, retries, fallbacks, and human-in-the-loop review. - Build evaluation loops to test output quality, consistency, hallucination risk, retrieval quality, and workflow reliability. - Collaborate with full-stack engineers to integrate AI workflows into product experiences. - Monitor and improve cost, latency, observability, reliability, and user trust. - Prototype quickly, then turn successful prototypes into maintainable product systems. **What You Bring ** - Experience building with LLM APIs and modern AI application patterns. - Strong understanding of agentic workflows: tool use, structured outputs, and multi-step task orchestration. - Practical experience in ML pipelines such as embeddings, similarity search, classification, evaluation metrics, and data quality measurement. - Experience with RAG, embeddings, vector databases, or retrieval systems — but not only as a basic chatbot pattern. - Ability to design workflows with state, constraints, fallback paths, human review points, and observability. - Ability to evaluate AI outputs using examples, rubrics, regression tests, human feedback, output validation, or automated checks. - 6+ years of professional software engineering experience in Python, TypeScript, APIs, databases, and backend application logic. - Comfort building AI systems that connect to product workflows, internal tools, data sources, and external services. - Product judgment: you care whether AI is useful, understandable, reliable, and trustworthy to users. - Clear communication and comfort working in an early-stage, async-first startup environment. Strong signals **We are especially interested in candidates who have built AI systems where: ** - The LLM calls tools or APIs to complete a task. - The workflow has multiple steps, not one prompt-response interaction. - The system uses structured data, schemas, or validated outputs. - The pipeline collects, cleans, classifies, ranks, or scores real-world data. - Outputs are traceable, reviewable, and measurable. - The system handles uncertainty, missing context, or failure cases. - There is an evaluation or feedback loop. - A human can review, approve, or correct important outputs. - The AI feature was used inside a real product, internal tool, customer workflow, or production-like environment. **Nice to have ** - Experience with agent frameworks, workflow orchestration, MCP/tool-use patterns, or multi-model systems. - Experience building AI copilots, research agents, workflow agents, document agents, or task automation systems. - Experience with evaluation frameworks, observability, prompt/version management, or AI quality monitoring. - Experience with entity resolution, knowledge graphs, search/ranking, data enrichment, or intelligence pipelines. - Experience with hiring, talent, HR tech, recruiting, marketplace, or workflow-heavy SaaS products. - Prior startup, open-source, side project, or internal tool experience that shows end-to-end AI product building. - What this role is not Foundation model training. Research science. Prompt engineering alone. A basic RAG chatbot. Notebook demos. We are looking for someone who can apply LLMs, retrieval, lightweight ML, data pipelines, and agentic workflow design to build reliable product intelligence that people depend on. THIS IS THE RIGHT ROLE FOR YOU IF… - You want your name on the architecture. Not on a Jira ticket. You have watched other people make the foundational calls and thought: I would have done that better. - You are comfortable with ambiguity — genuinely, not as an interview answer. You will have a spec. Some weeks you will decide what to build as well as how. - You are ambitious about outcome, not title. The title is negotiable and will grow. What is not negotiable is that you are accountable for whether this layer works. - You find the math interesting. You do not need to know what a Fisher information matrix is. - - - You need to be the kind of person who would look it up. - You can carry the equity-only period. Financially and psychologically. Six months is a long time to build on conviction. IT IS THE WRONG ROLE IF… - You want work-life balance right now. Later, yes. In the pre-launch phase, no. We would rather say that plainly than have you discover it in month two. - You are optimizing for total cash compensation over the next 18 months. A senior role at a funded company will pay you more, sooner, with less risk. That is a completely rational choice. - You want to specialize narrowly. You will touch pipelines, evaluation, prompt architecture, data quality, and product integration — often in the same week.