Job description

INTRODUCING KEPLER THE PROBLEM High-stakes industries are falling behind on AI adoption. Their workflows can’t afford wrong answers. And AI can’t be trusted to give right ones because of hallucinations. The barrier isn’t that the models aren’t smart enough. It’s that no one can verify what they produce. The fix isn’t a better model, it’s a trust layer: every output traceable, every calculation auditable, every answer reproducible. WHAT KEPLER IS Kepler is the agent harness - the infrastructure layer that wraps around AI models to make their outputs reliable, traceable, and verifiable. The model is a replaceable component. The harness is the product. In Kepler's architecture, the LLM orchestrates - it decides what data to gather, what to compute, how to structure the output. But every actual data point, every extracted value, every calculation flows through deterministic code pipelines. The LLM never touches the data itself. Every value carries provenance metadata back to its exact source. Every computation is auditable and reproducible. Verification loops cross-check outputs before users ever see them. We started in finance because the stakes are highest and the tolerance for error is zero. We’ve built a finance research product that lets analysts supercharge their workflow: pulling comparables, building models and researching filings. No more double-checking every number AI spits out. Every number tracing back to the source, every time. But the architecture - provenance, deterministic computation, verification - applies anywhere trust in AI output matters: chemicals, legal, healthcare. Models are commoditizing fast. The trust layer is what's missing and the market is massive. THE TEAM The founding team spent a combined 40+ years at Palantir building the type of large-scale data infrastructure that Kepler requires. Our founding engineers led Foundry's core systems - Ontology, Fusion, Workshop, FoundryML, created Palantir's first AI platform and scaled data products at Meta to 1B+ users. We’ve paired this deep technical foundation with a repeat founder profile. Our CEO built and scaled a data company to $15M ARR before successfully selling it. He then became Citadel's first Head of Business Engineering, experiencing first hand the problems we are now solving. We have a team who’ve been on both sides: building systems like this at massive scale and selling it into the buyers who need it most. We’re backed by investors who built the modern AI and data stacks, plus the builders of iconic commercial businesses. This includes founders of OpenAI, Meta AI Research, MotherDuck, dbt Labs and Square as well as PebbleBed, Company Ventures and Mantis VC firms. THE ROLE WHAT YOU'LL OWN You'll own the models inside Kepler's AI research platform: which model runs each task, when a fine-tuned model beats a frontier one, and the training, evaluation, and extraction systems that make every workflow powerful. Model-agnostic by design doesn't mean the model doesn't matter. It means model choice is a permanent engineering problem, and it's yours. The models you choose and tune sit inside a product financial professionals rely on for million-dollar decisions. This role is for engineers who want to build foundational technology at the intersection of AI and finance, where your code directly impacts how clients make critical business decisions. In the first few weeks you might: - Fine-tune a small model on a high-volume extraction task (footnote tables in 10-Ks, IR decks) and show it beats the frontier model we use today on accuracy, cost, and latency. - Build an eval harness that scores agent research runs end to end (does every number trace, does every citation resolve) and wire it into CI so regressions get caught before analysts see them. - Redesign model routing across a workflow: a frontier model where the reasoning is hard, cheaper or fine-tuned models for high-volume extraction and verification steps, with evals proving nothing got worse. - Take a workflow that succeeds 80% of the time and systematically find the other 20%: better tools, tighter verification rules, different context, a fine-tune, or a different model entirely. In the longer term, you’ll be given ownership of whole functional areas, from extending our platform to a new industry to leading new architecture as our infrastructure scales. You'll consistently own systems end-to-end. In a small team, there's nobody to hand things off to. HOW WE WORK We’re a close team, working together in an office in New York. We use AI tools heavily - Cursor, Claude Code, whatever makes us faster. Fluency is assumed. Our users are analysts at firms where a wrong number costs real money. The feedback loop on what you ship is hours, not quarters. The pace is startup-fast but the engineering bar is high. We care about getting things right, not just getting things out. If you've worked somewhere that moves fast but ships broken software, this is different. If you've worked somewhere that's rigorous but slow, this is also different. The team has strong backgrounds and low ego. We expect everyone to roll up their sleeves and handle the unglamorous problems: the weird regressions, the subtle bugs, the last minute debugging session before a demo. We move as a team, not as a collection of individuals. WHO YOU ARE You've shipped production systems and you care about whether they're correct - not just whether they work on the happy path. You think about failure modes before someone asks you to. You're comfortable in a codebase you didn't write, moving between a fine-tuning run and the orchestrator code that serves the result in the same day. You're drawn to early-stage not for the title but because you want your work visible in the product, not abstracted behind three layers of management. From the technical side: - 5+ years building production software. No upper limit, comp scales with experience. - You've shipped ML systems to production (fine-tuning, agents, retrieval, structured extraction) and you know what breaks between a demo and a product. - You treat evals as engineering: you build the measurement before the feature, and you don't call something better until the numbers say so. - Strong general engineer, whatever your path into ML (research, ML infra, product). Our backend is Rust, but we don't require Rust experience. We believe strong engineering fundamentals and experience in other languages is what matters. - You’re a quick learner and are as comfortable in a codebase you wrote as one you’re reading for the first time. From the personal side: - You care what the analyst does with what you shipped, not whether the code was clever. - You’d rather fix something than file a ticket about it. - You’ll tell someone their design has a flaw before the PR goes in, not after. - You communicate before it’s a problem, and when a teammate needs something from you, they don’t have to ask twice. - You know what it feels like when the plan changes twice in a day and the work still has to ship. Don't check every box? Apply anyway. We prioritize problem-solving ability, systems thinking, and drive to build transformative agentic infrastructure. OUR TECHNICAL STACK - Backend: Rust - agent orchestration, data extraction, computation pipelines. - Frontend: TypeScript, React - the analyst workspace and verification interfaces. - Data: PostgreSQL, plus direct integrations with official data sources. - Infra: AWS. - AI: Model-agnostic by design. We currently use Claude and GPT. The model is the replaceable part. MENTORSHIP & GROWTH You'll be directly mentored by engineers who built Palantir's core systems. Expect: - Weekly 1:1s with senior engineers who've built systems at Palantir and Meta scale. - Deep architectural reviews and guidance on system design. - Clear growth path toward technical leadership and system ownership. - Learn by building production systems that power real financial research. WORKING AT KEPLER OUR BENEFITS - 100% covered top-of-the-line medical, dental, and vision insurance for employees and their families. HSA maxed by the company to the IRS limit. - Automatic coverage for life, AD&D, and disability insurance. - Daily lunch in office. - Unlimited PTO policy. - Development environment budget - latest MacBook Pro, multiple monitors, ergonomic setup, and any development tools you need. - "Build anything" budget - dedicated funding for whatever tools, libraries, datasets, or infrastructure you need to solve technical challenges, no questions asked. - Learning budget - attend any conference, course, or program that makes you better at what we're building. OUR OPERATING PRINCIPLES - Trust as the Default: People do their best work when confidence is mutual. We show our work, keep our promises, and flag risks before they bite. Trust isn't an aspiration - it's the baseline. - Forward-Deployed with Product DNA: We own customer outcomes while building a product company. We don't win if they don't win. - Extreme Ownership: If you notice a problem, you own it by making sure it doesn’t fall through the cracks. Authority comes from initiative, not job titles. Once you step up, you're accountable for the outcome. - Production-First Engineering: We design for critical workloads from day one. Durable execution, blue/green deploys, automated rollbacks, continuous delivery with end-to-end observability. - Communicate with Intent: Great work disappears without great communication. We push information to the people who need it, when they need it. Silence is never the safe choice. - Earn it Every Day: Your work speaks for itself. We create an environment where the best idea wins, the strongest work gets recognized, and everyone is held to the same high standard. - Keep Raising the Bar: Great teams compound. Every hire raises the bar, every win gets named, every person gets the tools and runway to grow. Kepler is an Equal Opportunity Employer and prohibits discrimination and harassment of any kind. We are committed to the principle of equal employment opportunity for all employees and to providing employees with a work environment free of discrimination and harassment.

Requirements

Must-haves

Kepler is the agent harness - the infrastructure layer that wraps around AI models to make their outputs reliable, traceable, and verifiable. The model is a replaceable component. The harness is the product.
This role is for engineers who want to build foundational technology at the intersection of AI and finance, where your code directly impacts how clients make critical business decisions.
Fine-tune a small model on a high-volume extraction task (footnote tables in 10-Ks, IR decks) and show it beats the frontier model we use today on accuracy, cost, and latency.
Build an eval harness that scores agent research runs end to end (does every number trace, does every citation resolve) and wire it into CI so regressions get caught before analysts see them.
Redesign model routing across a workflow: a frontier model where the reasoning is hard, cheaper or fine-tuned models for high-volume extraction and verification steps, with evals proving nothing got worse.
Take a workflow that succeeds 80% of the time and systematically find the other 20%: better tools, tighter verification rules, different context, a fine-tune, or a different model entirely.
In the longer term, you’ll be given ownership of whole functional areas, from extending our platform to a new industry to leading new architecture as our infrastructure scales.
You'll consistently own systems end-to-end. In a small team, there's nobody to hand things off to.
You've shipped production systems and you care about whether they're correct - not just whether they work on the happy path. You think about failure modes before someone asks you to.
5+ years building production software. No upper limit, comp scales with experience.
You've shipped ML systems to production (fine-tuning, agents, retrieval, structured extraction) and you know what breaks between a demo and a product.
You treat evals as engineering: you build the measurement before the feature, and you don't call something better until the numbers say so.
You’re a quick learner and are as comfortable in a codebase you wrote as one you’re reading for the first time.
You care what the analyst does with what you shipped, not whether the code was clever.
You’d rather fix something than file a ticket about it.
You’ll tell someone their design has a flaw before the PR goes in, not after.
You communicate before it’s a problem, and when a teammate needs something from you, they don’t have to ask twice.
You know what it feels like when the plan changes twice in a day and the work still has to ship.
Don't check every box? Apply anyway. We prioritize problem-solving ability, systems thinking, and drive to build transformative agentic infrastructure.
Backend: Rust - agent orchestration, data extraction, computation pipelines.
Frontend: TypeScript, React - the analyst workspace and verification interfaces.
Data: PostgreSQL, plus direct integrations with official data sources.
AI: Model-agnostic by design. We currently use Claude and GPT. The model is the replaceable part.
You'll be directly mentored by engineers who built Palantir's core systems. Expect:

Nice-to-haves

None flagged.

Application prep

Use the role’s language and application details while tailoring your materials.

Keywords to mirror

aillmmlrusttypescriptreactaws

Platform: ashby
Effort: quick

Benefits

Benefits detected in the listing.

Dental

Vision

Pto

Learning Budget