Über diese Staff Applied AI Engineer - Kalepa for Underwriting Stelle bei Kalepa
About Kalepa:
Insurance is a $7T global industry built on professional labor: underwriting, claims, pricing, operations, and distribution. Kalepa is building AI that can perform that work.
We create professional-grade AI systems for insurers: software that can read, reason, decide, and act inside real workflows and operate over long-time horizons. Kalepa begins by augmenting professionals with the tools to succeed at their jobs. That earns us the right to learn from real decisions, capture decision traces, and automate what can be automated successfully over time.
Our products are already in production with leading insurers, driving measurable outcomes and expanding across functions. The opportunity is far broader than any single workflow: to build the AI operating system for decision-making across insurance. For people who want to work on AI that matters, in a sector where correctness, trust, and real-world impact count, Kalepa is just getting started.
Kalepa is backed by leading investors such as IA Ventures and Inspired Capital, and our team brings experience from Meta, Palantir, Block, Affirm, Google, Amazon, McKinsey, Mastercard, and Uber.
Our Values (This is important):
Our culture is shaped by everyday decisions. Our values help us make them thoughtfully and consistently, especially when tradeoffs are difficult. We learn from mistakes so our good decisions outweigh our bad ones and their benefits compound over time.
- Excellence - We take pride in our craft and expect excellent outcomes in everything we do. We set high standards, verify our work, and own the result. We make deliberate tradeoffs, including rethinking an approach, redoing work, or investing more time.
- Pursuit of the Truth - We seek the truth, even when it is uncomfortable. We question assumptions, give candid feedback, and change our minds when evidence warrants it. We work with customers to understand their challenges and what solving them requires.
- Ensuring Customer Success - We take responsibility for delivering results that matter to our customers. We measure success through improvements in their growth, profitability, and efficiency. We challenge our approach and customers’ requests when necessary, remaining accountable for the outcome.
- First Principles Thinking - We reason from a problem’s underlying facts, objectives, and constraints. We distinguish what is necessary from what is familiar and question assumptions that limit possible solutions. We test our reasoning against reality and address root causes with approaches that scale without unnecessary complexity.
In addition to our NY office, we have remote employees speaking 10+ languages across the globe. But we invest in bringing our people together both for regional meetups and global offsites (2021 - Playa Del Carmen | 2022 - Rome | 2023 - Buenos Aires | 2024 - Lisbon | 2025 - Cartagena | 2026 - Tbilisi).
Kalepa’s culture isn’t for everyone, and that’s ok. But for the people who are a fit, they can’t imagine working elsewhere.
About the role
We are looking for a Staff Applied AI Engineer to set the technical direction for how Kalepa turns unstructured insurance data into structured, explainable facts at scale. Kalepa's platform ingests billions of highly unstructured data points every day, and the LLM-based extraction, classification, and matching systems that make sense of them sit at the center of the product. You'll own the architecture of those systems, the evaluation infrastructure that governs them, and the model strategy behind them, making the calls under ambiguity about which approaches, models, and vendors Kalepa bets on.
You'll partner with product, engineering, and business leadership to tie model outputs directly to customer value, and you'll be the go-to resource on applied AI for engineers and non-engineers alike. This is a problem the industry hasn't solved effectively yet. We're a small team where individual contributions have an outsized impact on both our technology and our business.
Salary range
We differentiate our North America salary ranges using two geographical zones:
Zone 1 (New York and San Francisco) - $270,000 to $345,000*
Zone 2 (all other locations in the US and Canada) - $243,000 to $310,500*
Location
We are primarily a remote company with employees across the US, Canada, and Europe. Our HQ is in New York city.
About the team
This team owns Kalepa’s proprietary risk representation layer and intelligent model-driven insights to drive user actions. It converts Kalepa’s past submissions and syncs into a durable data moat.
Ownership areas: risk representation, winnability model, future models such as loss prediction, portfolio optimizers, and more.
What you'll do
- Lead the technical, data, and project aspects of Kalepa's key applied AI initiatives: document extraction, classification, entity matching, and the evaluation platform behind them.
- Set the technical direction for how Kalepa builds and operates LLM-based systems at scale: accuracy, cost, latency, reliability, and explainability.
- Lead architectural decisions across data science and the backend; be a strong systems designer.
- Own the evaluation infrastructure: benchmark datasets, regression suites, validation strategies, and the standards every model change must meet before release.
- Deliver major product and data science innovations in each release, and consistently tie model outputs to customer value.
- Lead model and vendor strategy, including when to adopt new models, when to fine-tune, and when to replace a pipeline.
- Partner with product and business owners to drive firm-wide decisions, pragmatically weighing data, resources, and tradeoffs.
- Proactively manage technical debt across the AI stack; lead code reviews and the adoption of best practices for development and deployment.
- Act as an ambassador for Kalepa internally and externally, including in recruiting and training.
- Use AI tooling deliberately to multiply your own impact and that of the team.
About you
- 8+ years of engineering experience, with 5+ years building and operating ML or LLM systems in production.
- Have delivered as the technical lead on at least one major applied AI initiative or product.
- Deep expertise in LLM-based systems: structured extraction, classification, retrieval and embeddings, evaluation, and the engineering required to run them reliably and economically at scale.
- Mastery of applied machine learning, NLP, and statistics, and of Python and its data and ML libraries.
- Experience with AWS and serverless technologies (Lambda, EC2), asynchronous workflows, and cloud environments.
- Consistently a go-to resource for engineering, product, and business staff in technical matters.
- You take full ownership of your work and drive initiatives to completion with minimal oversight.
- You communicate proactively and persuasively with engineers, product, and leadership.
- You thrive on experimentation and are excited by problems others are struggling with.
Bonus points for
- Experience with document AI at scale: OCR, layout understanding, long-document and multi-page extraction.
- Experience fine-tuning or distilling models for production use.
- Experience building evaluation platforms or ML infrastructure used by other teams.
- Familiarity with insurance, underwriting, or submission documents such as ACORD forms and loss runs.
What you’ll get
- Competitive salary (based on experience level)*.
- Significant equity options package.
- 20 days of PTO a year.
- Global team offsites (2024 - Lisbon, 2025 - Cartagena, 2026 - Tbilisi).
- Healthy living/gym stipend.
- Mobile phone bill stipend.
- Continuing education credits.
For US employees only:
- 100% covered medical, 100% covered vision and dental for individuals and families.
- 401(k) plan with employer contribution (regardless of employee contribution)
*The salary range listed is an estimate and will vary based on a variety of factors. Final compensation will be determined during the offer stage based on relevant experience, performance during the interview process, and geographic location, and may therefore differ from the posted range.
