AI/ML Engineer

AI/ML Engineer at Keeper | Y Combinator's Work at a Startup Work at a StartupStartup JobsInternshipsUpcoming EventsHow it WorksLog InAI/ML Engineer at Keeper (W19)$145K - $165K • 0.24% - 0.32%AI tax filing for complex returnsSan Francisco, CA, USFull-timeUS citizen/visa only1+ yearsApply nowAbout Keeper Keeper is AI-powered tax filing software that replicates the experience of going to an accountant. Just drop off your forms, answer a few questions, and let Keeper do the rest! About the roleSkills: Python, Machine Learning COME BUILD THE FIRST AI-NATIVE TAX SOFTWARE Tax software hasn't had a real shakeup in decades. We're changing that. [Keeper](https://keepertax.notion.site/Working-at-Keeper-98a6be20e7e44922be6291f5513af40d) is building the future of tax filing — not by bolting AI onto a legacy product, but by rebuilding the entire experience from the ground up around what AI makes possible today. And we've been doing this longer than almost anyone. We were using OpenAI's language models to clean transaction data long before ChatGPT existed. Our in-house document parsing transformers are still faster and more accurate than state-of-the-art multimodal models. Our [AI tax assistant](https://www.keepertax.com/research/lessons-learned-from-supercharging-live-tax-assistance-with-llms) consistently [outperforms every competitor](https://www.keepertax.com/research/keepers-chatbot-vs-industry-incumbents) we've benchmarked it against. We're not chasing the AI wave; we've been ahead of it from day one. Now we're entering the most exciting phase yet: building features that literally weren't possible a year ago. Think holistic tax intelligence extracted from uploaded documents, agentic tax filing workflows, and next-gen evals to make it all production-ready. This is the rare opportunity to work on AI that ships to real users solving real problems — not a research project, not a demo. Us, in numbers: - [YC-backed](https://keepertax.notion.site/Working-at-Keeper-98a6be20e7e44922be6291f5513af40d), $15M raised (Series A), 20 full-time employees - 40K paying users ($199–$399/year) - 20K reviews averaging 4.8 stars across the [App Store](https://apps.apple.com/us/app/keeper-taxes-made-magical/id1471090941), [Play Store](https://play.google.com/store/apps/details?id=com.keepertax&hl=en_US&gl=US&pli=1), and [TrustPilot](https://www.trustpilot.com/review/keepertax.com) --- THE ROLE We're hiring an AI/ML engineer to design, build, and ship the core intelligence behind Keeper's product. You'll work across the full stack of modern AI — from model development to prompt engineering to production infrastructure — with an intense focus on turning capabilities into features users love. What you'll work on: - Agentic tax filing — orchestrating LLM-powered agents that can reason about a user's full tax picture and take action on their behalf - Document intelligence — extending our [best-in-class transformer parser](https://www.keepertax.com/research/fine-tuning-a-transformer-model-for-document-data-extraction) to extract holistic tax context from uploaded documents - AI tax assistant — pushing our already [industry-leading](https://www.keepertax.com/research/keepers-chatbot-vs-industry-incumbents) assistant further with better retrieval, richer tool use, and deeper tax reasoning - Evals and reliability — building the testing and evaluation infrastructure that lets us ship AI features with confidence - Core ML models — iterating on our NLP and classification models for [transaction processing and write-off detection](https://www.keepertax.com/research/automatic-write-off-detection) --- WHAT WE'RE LOOKING FOR - Deep fluency in the modern AI toolkit — LLMs, RAG, agents, fine-tuning, prompt engineering, evals. You stay close to the frontier and have strong intuitions about when to use what. - Strong engineering chops — highly proficient in Python and comfortable picking up whatever the job demands — JS/TS, GCP infrastructure, new ML tooling, you name it. The stack is evolving fast; what matters is that you write clean, production-grade code and learn new tools quickly. - Obsessively product-minded — you think in terms of user experiences, not model metrics. You want to build features that feel so smart they're magical. - Fast and iterative — you design smart experiments, analyze results quickly, and close the loop. You'd rather ship and learn than deliberate endlessly. - Ownership-oriented — you take end-to-end responsibility for what you build, from prototype to production to on-call. - Collaborative — you partner closely with engineers, product managers, and tax experts across the team. Tax domain experience is a plus but not required. What matters most is passion for building great AI-powered products. --- LOCATION Hybrid in San Francisco. We're looking for someone who wants to work one or more days per week out of our office in the Financial District. VISA REQUIREMENTS Applicants must be a U.S. citizen