ML Engineer - NLP

Full Description

About The Role

Autopoiesis Sciences is an applied AI lab based in San Francisco, California on a mission to accelerate scientific discovery across every discipline.

We're building AI systems from first principles, developing novel architectures that combine calibrated confidence estimation with reinforcement learning from experimental outcomes.

Our approach learns optimal research strategies directly from real-world scientific practice, training on experimental results, publication impact, and validated discoveries rather than static knowledge alone.

By closing the loop between AI hypothesis generation and experimental validation, we're creating systems that know what they don't know and improve through ground truth from nature itself.

Backed by Informed Ventures, Alpaca VC, Cross Atlantic Angels, Adam Grosser, Ally Warson, Mike Mahlkow, Hiro Mizushima, and others, we're preparing to release our systems to the research community.

We invite you to join us in building the future of scientific intelligence.

Location

Autopoiesis Sciences is headquartered in San Francisco, and we're expanding our engineering team to Toronto to tap into Canada's exceptional AI and engineering talent ecosystem.

This is a fully remote position for candidates based in the Greater Toronto Area.

Who We're Looking For

The breakthroughs that actually change the world come from people who refuse to accept "impossible" as an answer.

We want individuals who approach their work with intellectual humility.

People who know that despite all the progress in machine learning, we're still scratching the surface of what's possible.

You should be comfortable admitting what you don't know, but absolutely relentless about figuring it out.

The kind of person who, when faced with a problem everyone says can't be solved, treats that as the starting point rather than the conclusion.

This isn't a role for people who need their hands held or who expect clear answers to ambiguous problems.

Scientific superintelligence won't be built by following established playbooks.

It requires questioning assumptions, rebuilding from first principles when necessary, and maintaining conviction even when the path forward is unclear.

We need people who derive energy from hard problems and won't settle for "good enough" when breakthrough solutions are possible.

If you're someone who sees the current state of machine learning and thinks "we can do better," who believes the most important work is still ahead of us, and who's willing to put in whatever it takes to get there, we want to hear from you.

Research Areas of Interest

• LLM training and fine-tuning for domain-specific scientific reasoning

• RL methods for improving language model reasoning and decision-making

• Scientific literature understanding and knowledge extraction

• Multi-document synthesis and claim verification

• Retrieval-augmented generation for scientific knowledge

• Alignment and RLHF for scientific accuracy and rigor

• Evaluation frameworks for scientific language understanding

Autopoiesis Sciences is seeking exceptional ML Engineers with expertise in language models and reinforcement learning to help build AI systems that can understand, reason about, and generate scientific knowledge.

As an ML Engineer, you will develop and optimize large language models that form the cognitive core of our autonomous research systems, enabling them to comprehend complex scientific literature, reason through multi-step problems, and communicate discoveries effectively.

You'll apply RL techniques to improve reasoning capabilities and train models that learn from experimental feedback and scientific validation.

While language understanding is central to this role, you'll work on systems that integrate LLMs with broader AI capabilities including autonomous planning, hypothesis generation, and experimental design.

This role offers the opportunity to push the boundaries of what language models can achieve when combined with reinforcement learning in scientific contexts.

Responsibilities

• Design and implement large-scale LLM training and fine-tuning pipelines for scientific domains, including methods for domain adaptation, instruction tuning, RLHF, and RL-based reasoning improvements

• Develop RL algorithms that improve language model reasoning, including reward modeling from scientific feedback, process supervision, and outcome-based learning

• Build systems for scientific literature understanding, including information extraction, claim verification, citation analysis, and cross-document reasoning

• Create retrieval systems and knowledge bases that enable LLMs to access and synthesize information from vast scientific corpora

• Develop evaluation frameworks that measure language models' scientific reasoning capabilities, accuracy, and adherence to proper epistemic standards

• Optimize inference systems for efficient deployment of large language models in production research workflows

• Collaborate with research scientists to encode domain expertise into LLM training objectives and RL reward functions

• Work on the full lifecycle of language and RL systems from research and prototyping through production deployment

Qualifications

• Bachelor's degree or higher in Computer Science, Machine Learning, Computational Linguistics, or a related quantitative field

• 2-3 years of experience in machine learning engineering, LLMs, or related areas (internships, research positions, or industry experience)

• Strong programming skills in Python and deep experience with modern ML frameworks (PyTorch preferred, JAX or TensorFlow acceptable)

• Demonstrated expertise in transformer architectures, attention mechanisms, and language model training or fine-tuning

• Experience working with large language models (training, fine-tuning, RLHF, or deployment at scale)

• Solid understanding of reinforcement learning fundamentals and their application to language models

• Strong grasp of information retrieval, text processing pipelines, and evaluation metrics for language systems

• Proven ability to implement research papers and adapt techniques to new domains

• Experience with distributed training, model optimization, or MLOps practices

• Bonus: Published papers in ML/NLP conferences, contributions to major LLM projects or libraries, experience with RL for LLMs (RLHF, PPO, DPO, etc.), scientific text processing experience, or background in a scientific domain

Compensation And Benefits

• Competitive compensation, equity, and benefits

• Collaborate in building a thoughtful, mission-driven team culture focused on scientific discovery

• Frequent team events, dinners, off-sites, and gatherings

• Compensation packages are highly variable based on experience, expertise, and contribution potential

Get To Know Us

We're a dedicated team of scientists, researchers, and engineers who believe deeply in the transformative potential of our work.

You'll work closely with our leadership team including Chief Executive Officer Joseph Reth (Attended University for CS at 14, former DARPA), Chief Business Officer Dr.

Eike Gerhardt (University of Tübingen PhD, University of Tübingen staff), and Chief Scientist Dr.

Larry Callahan (University of Chicago PhD, former FDA, former NIH).

Application Process: Due to the high volume of automated applications, we only accept applications through LinkedIn as it helps us connect with genuine candidates.

We have systems in place to detect automated submissions and strongly advise against using bots or automated tools in your application process.

Please be human in your approach.

Equal Opportunity: Autopoiesis Sciences, Inc. is an equal opportunity employer committed to diversity and inclusion. We welcome applications from all qualified candidates regardless of race, gender, age, religion, sexual orientation, or any other legally protected characteristics.