Machine Learning Engineer

Kepler • New York, NY

29d • $200,000 - $280,000 • Onsite

About The Position

Kepler is building 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, but the harness is the product. In Kepler's architecture, the LLM orchestrates, deciding what data to gather, what to compute, and how to structure the output. However, every actual data point, extracted value, and calculation flows through deterministic code pipelines, ensuring the LLM never directly touches the data. Every value carries provenance metadata back to its exact source, every computation is auditable and reproducible, and verification loops cross-check outputs before users see them. Initially focused on finance due to the high stakes and zero tolerance for error, Kepler has developed a finance research product enabling analysts to supercharge their workflows by pulling comparables, building models, and researching filings with confidence in data accuracy. The core architecture of provenance, deterministic computation, and verification is applicable to any industry where trust in AI output is critical, such as chemicals, legal, and healthcare. As models commoditize, the trust layer is the key differentiator in a massive market.

Requirements

• 5+ years building production software.
• Shipped ML systems to production (fine-tuning, agents, retrieval, structured extraction) and understand the challenges of moving from demo to product.
• Treat evals as engineering: build measurement before the feature and rely on data to confirm improvements.
• Strong general engineering fundamentals, with experience in other languages being valued over specific Rust experience.
• Quick learner, comfortable in both self-written and unfamiliar codebases.
• Focus on the user's needs and the correctness of the shipped product over clever code.
• Proactive in fixing issues rather than just reporting them.
• Willingness to provide constructive feedback on designs before PR submission.
• Effective communication, proactively sharing information and responding promptly when teammates need assistance.
• Adaptability to changing plans and ability to ship work under pressure.
• Prioritize problem-solving ability, systems thinking, and drive to build transformative agentic infrastructure.

Nice To Haves

• Rust experience (though not required).

Responsibilities

• Own the models inside Kepler's AI research platform, determining which model runs each task, when a fine-tuned model outperforms a frontier model, and managing the training, evaluation, and extraction systems.
• Make model choice a permanent engineering problem, impacting a product financial professionals rely on for million-dollar decisions.
• Build foundational technology at the intersection of AI and finance, where code directly impacts critical business decisions.
• Fine-tune models for specific tasks (e.g., footnote table extraction from 10-Ks, IR deck analysis) to improve accuracy, cost, and latency compared to existing frontier models.
• Develop an evaluation harness to score agent research runs end-to-end, ensuring data traceability and citation resolution, and integrate it into CI to catch regressions.
• Redesign model routing within workflows, utilizing frontier models for complex reasoning and cheaper or fine-tuned models for high-volume extraction and verification, with rigorous evals to ensure no degradation in performance.
• Systematically identify and resolve issues in workflows that succeed only partially (e.g., 80% success rate) by improving tools, verification rules, context, fine-tuning, or model selection.
• Take ownership of entire functional areas, potentially extending the platform to new industries or leading new architectural developments as infrastructure scales.
• Own systems end-to-end, as there is no one to hand off tasks to in a small team.

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 needed).
• "Build anything" budget (dedicated funding for tools, libraries, datasets, or infrastructure).
• Learning budget (attend any conference, course, or program).

What This Job Offers

• Job Type: Full-time
• Career Level: Senior
• Education Level: No Education Listed