Founding Engineer

Goliath Partners Inc.; United States; Remote; USD 260,000 - 400,000; Full time

Benefits offered by this job

Equity stake

Unlimited AI tools budget

Comprehensive benefits

Job summary

Goliath Partners Inc. seeks a founding RL engineer to own the end-to-end reinforcement learning loop for a YC-backed AI startup focused on drug discovery.

You’ll design RL environments, run post-training experiments, and define rewards enabling frontier medical models to reason over clinical and genomic data. Join as a founding team member to set technical direction, build scalable pipelines, and help hire.

Qualifications

• 1+ years of deeply technical work in reinforcement learning, agent environments, LM post-training, or related ML systems.
• Strong judgment about data quality: signal fidelity, coverage, and relevance, and what makes a training task meaningful.
• Ability to build scalable pipelines, debug large ML codebases, and move from research idea to working prototype.

Responsibilities

• Build RL environments end to end, including task design, action spaces, tool interfaces, verifiers, and evaluation harnesses.
• Run post-training experiments on language models and agents using SFT, RLVR, RLHF/RLAIF, and reward modeling.
• Train and evaluate multi-step agents that navigate complex patient histories and unstructured medical data, capturing what correct clinical reasoning looks like.
• Dig into model failures and use findings to improve data, rewards, and feedback signals.
• Help set the technical direction, culture, and hiring bar as a founding team member.

Skills

Reinforcement learning

Data quality

ML pipelines

Job description

Our client is a YC-backed AI startup working to transform drug discovery, with a mission as ambitious as it gets: helping cure cancer. They’re building the foundational data layer and the pre- and post-training simulation environments where the next generation of medical AI models will learn.

They’ve raised a $10M pre-seed at a $100M valuation, backed by angel investors including executives from OpenAI and Anthropic. Now they’re assembling the founding team, and this is one of the first seats at the table.

You’ll own the full reinforcement learning loop and build the evaluation and reward layer that teaches frontier models to reason correctly about real-world clinical and genomic data. In healthcare, the difference between a model that sounds right and one that is right matters enormously, and you’ll be the one defining that difference.

What you’ll do

• Build RL environments end to end, including task design, action spaces, tool interfaces, verifiers, and evaluation harnesses
• Run post-training experiments on language models and agents using SFT, RLVR, RLHF/RLAIF, and reward modeling
• Train and evaluate multi-step agents that navigate complex patient histories and unstructured medical data, capturing what correct clinical reasoning actually looks like
• Dig into model failures and use what you find to improve the data, rewards, and feedback signals that shape what these systems learn
• Help set the technical direction, culture, and hiring bar as a founding team member

What you bring

• 1+ years of deeply technical work in reinforcement learning, agent environments, LM post-training, or related ML systems, whether in industry, research, or serious independent projects
• Strong judgment about data quality: you can assess signal fidelity, coverage, and relevance, and you know what makes a training task meaningful
• The ability to build scalable pipelines, debug complex training runs in large ML codebases, and move fast from research idea to working prototype

Nice to have

• Experience with healthcare, clinical, biological, or genomic data (not required; you’ll pick up the domain quickly alongside the team)
• Publications, open-source work, or projects in RL, evals, or post-training
• Prior early-stage startup experience

Why this role

• A mission that matters: your work goes directly toward faster drug discovery and better cancer treatment
• Founding-team ownership: shape the core technology and the company itself from day one
• Serious backing: YC, a $10M pre-seed at a $100M valuation, and angels from OpenAI and Anthropic leadership
• Frontier problems: RL environments and reward design for medical AI is one of the least solved and most important areas in the field
• Top-tier compensation: up to $400k base, up to 1.0% equity, comprehensive benefits, and an unlimited AI tools budget