Tacit Labs is building data, evaluations, and research environments unlocking scientific discovery capabilities for frontier models. Our goal is to help AI systems make progress on real, long-horizon problems in biology, chemistry, and drug development.

We’re looking for a Research Engineer to help build Tacit’s scientific agent environments: integrated environments where language-model agents can use tools, inspect data, make decisions, and solve increasingly complex scientific tasks over many steps.

This role is ideal for someone who has worked on RL environments, agent systems, language models, or scientific tooling, and wants to build the infrastructure that helps frontier models learn to reason and act in scientific domains.

What you’ll do

• Build long-horizon environments where agents can interact with tools, data sources, models, and scientific workflows.
• Design task interfaces, action spaces, observations, rewards, traces, and evaluation criteria for agentic scientific work.
• Integrate tools from biology, chemistry, and drug discovery into model-facing environments.
• Work on environments where agents may need to search literature, analyze data, call models, run computational tools, interpret results, and decide what to do next.
• Collaborate with research, engineering, product, and scientific experts to turn real scientific workflows into tractable environments.
• Build infrastructure for running, logging, debugging, and evaluating model behavior across multi-step tasks.
• Help define what makes an environment useful for training, evaluation, and model improvement.
• Work directly with founders and customers on some of Tacit’s most important technical bets.

What we’re looking for

• Strong engineering ability, especially in Python and modern ML tooling.
• Hands-on experience with language models, agents, tool use, evals, or RL environments.
• Familiarity with reinforcement learning paradigm
• Ability to build reliable systems around messy real-world workflows.
• Interest in biology, chemistry, drug discovery, or scientific AI. Some life-science background is strongly preferred.
• Comfort working across research and engineering: you can read papers, prototype quickly, and turn ideas into working systems.
• Strong debugging instincts and attention to detail.
• High agency and comfort operating in ambiguity.

You might be a fit if

• You’ve built RL environments, agent benchmarks, tool-use systems, coding environments, robotics/simulation environments, or eval infrastructure.
• You’ve worked with LLMs in a hands-on way, not just through prompting, but through system design, tool integration, evaluation, or training loops.
• You like turning complex workflows into clean abstractions.
• You’re interested in how agents can make progress on real scientific problems rather than toy tasks.
• You have enough biology or chemistry knowledge to be dangerous, and you’re excited to learn more.
• You want to work at the boundary of frontier AI, scientific reasoning, and practical engineering.

Tacit Labs is building data, evaluations, and research environments unlocking scientific discovery capabilities for frontier models. Our goal is to help AI systems make progress on real, long-horizon problems in biology, chemistry, and drug development.

We’re looking for a Researcher to help define, test, and improve long-horizon capabilities in language-model agents. This person will think deeply about how scientific tasks should be structured, how agent performance should be measured, and what kinds of data, environments, and feedback are needed to improve model behavior.

A core part of the role will be running RL experiments that dogfood Tacit’s evaluation and environment infrastructure. You will use our own platform to understand where agents fail, what kinds of supervision or environments help, and how to turn those findings into better products and research directions.

What you’ll do

• Design and run RL experiments on Tacit’s scientific evals and agent environments.
• Study long-horizon capabilities such as planning, tool use, error recovery, evidence integration, experiment design, and multi-step scientific reasoning.
• Develop hypotheses about what models need in order to improve: better tasks, better rewards, better data, better tools, better scaffolding, or better environment design.
• Work closely with our environment, eval, and platform teams to dogfood Tacit’s infrastructure and identify what needs to improve.
• Analyze model trajectories, failure modes, reward signals, task structure, and evaluation validity.
• Help define what makes a scientific environment useful for training, evaluation, and capability development.
• Partner with frontier AI labs as an extended part of their research teams, helping them understand what data, evals, and environments are needed to drive model progress.
• Translate research insights into concrete product, data, and environment requirements.

What we’re looking for

• Strong research taste and ability to think clearly about model capabilities, evaluation, and experimental design.
• Hands-on experience with language models, RL, post-training, agent systems, or evals.
• Ability to run experiments end-to-end: define the question, set up the run, inspect the outputs, analyze results, and decide what to try next.
• Interest in long-horizon reasoning, planning, tool use, and agent evaluation.
• Familiarity with reinforcement learning concepts such as rewards, rollouts, trajectories, credit assignment, exploration, and policy improvement.
• Enough engineering ability to work directly with experimental infrastructure, logs, model outputs, and environment code.
• Some understanding of biology, chemistry, drug discovery, or scientific workflows.
• Strong communication skills and comfort working directly with both internal teams and external research customers.

You might be a fit if

• You’ve worked on LLM post-training, RL environments, agent evals, tool-use benchmarks, or model behavior analysis.
• You like looking at messy model traces and figuring out what they reveal about the system.
• You are interested in the gap between benchmark performance and real long-horizon capability.
• You think carefully about what a task is actually testing.
• You have opinions about how to evaluate planning, scientific reasoning, and agentic workflows.
• You want to work closely with frontier AI labs on research problems that directly shape model development.
• You’re excited by the idea of using Tacit’s own infrastructure as a research testbed.

Tacit Labs is building data, evaluations, and research environments unlocking scientific discovery capabilities for frontier models. Our goal is to help AI systems make progress on real, long-horizon problems in biology, chemistry, and drug development. Our platform helps scientific experts design high-quality evaluation tasks, use AI models throughout the workflow, and turn expert judgment into structured datasets for training and evaluation.

We’re looking for a Product Engineer to build the internal eval platform used by hundreds of scientific experts, as well as Tacit’s research, engineering, and operations teams. This platform is where expert workflows happen: task design, rubric creation, model evaluation, review, QA, and delivery.

The ideal person is excited by ambitious internal tools, excellent user experience, and the challenge of making complex scientific work feel fast, structured, and intuitive.

What you’ll do

• Build the core platform scientific experts use to create, review, and refine evaluation tasks.
• Design product workflows that help experts move faster without lowering the quality bar.
• Integrate frontier AI models throughout the stack to assist with task generation, rubric design, answer analysis, QA, and review.
• Work closely with scientific experts to understand how they think, where they get stuck, and what tooling would make them dramatically more effective.
• Partner with research, engineering, product, and operations to turn delivery bottlenecks into better platform features.
• Build interfaces for complex structured data: prompts, answers, rubrics, reviews, model outputs, metadata, and customer deliverables.
• Create internal tooling that helps Tacit scale from bespoke expert workflows to repeatable, high-throughput eval production.
• Own features end-to-end, from product intuition and UX design through implementation, iteration, and deployment.

What we’re looking for

• Strong product engineering ability: you can talk to users, understand the real workflow, design a good solution, and ship it.
• Excellent taste in interfaces, workflows, and information architecture.
• Experience building complex web applications, internal tools, collaborative software, or evaluation platforms.
• Comfort integrating LLMs into real products, including prompting, model selection, structured outputs, review loops, and failure handling.
• Strong frontend skills, with enough backend ability to own features across the stack.
• High attention to detail and a bias toward simple, elegant systems.
• Ability to work in ambiguity and turn messy internal workflows into clear product surfaces.
• Interest in science, biology, drug discovery, or AI evaluation. Formal life-science experience is helpful but not required.

You might be a fit if

• You’ve built tools for knowledge work, creative work, research workflows, or expert review.
• You care about making complex work feel fluid and well-structured.
• You are highly opinionated about what “good” looks and feels like
• You’re excited by “tools for thought” applied to scientific reasoning and model evaluation.
• You want to work directly with users, ship quickly, and see your product immediately affect the quality and speed of real customer work.
• You’ve worked on an evals product, AI workflow product, internal platform, or expert-facing data product.

About the role

Tacit Labs is building data, evaluations, and research environments unlocking scientific discovery capabilities for frontier models. Our goal is to help AI systems make progress on real, long-horizon problems in biology, chemistry, and drug development. We work with expert scientists to create high-quality tasks, datasets, and workflows that help AI systems reason through real problems in biology, chemistry, and drug development.

We’re looking for an Project Manager to own customer projects end-to-end. This person will be responsible for making sure our most important projects are delivered on time, at a high quality bar, and with the level of rigor expected by the top AI labs in the world.

This is an unusually high-impact role for someone early in their career. You will work directly with the founders on Tacit’s most critical customer work, sit close to product and research decisions, and help define the operating system for a fast-growing company.

What you’ll do

• Own customer projects from kickoff through delivery, making sure timelines, quality standards, and deliverables are clear and met.
• Coordinate across expert contributors, product, engineering, research, and GTM to keep projects moving.
• Track project status, identify blockers, follow up aggressively, and make sure nothing falls through the cracks.
• Review work for completeness, consistency, formatting, and quality before it reaches customers.
• Help translate customer needs into internal workflows, product requirements, and roadmap priorities.
• Build and improve operational processes as we scale from bespoke projects to repeatable delivery systems.
• Jump into whatever is needed — project tracking, customer updates, QA, contributor coordination, documentation, data review, or workflow design.

What we’re looking for

• 1–3 years of experience, or exceptional early-career talent with strong evidence of ownership.
• Extremely organized, detail-oriented, and reliable under tight deadlines.
• Comfortable operating in ambiguity and turning messy situations into clear plans.
• High agency: you notice what needs to be done and do it without waiting to be asked.
• Strong written communication and ability to make complex work legible to customers and internal teams.
• Willingness to be very hands-on and deep in the details
• Interest in biology, life sciences, drug discovery, or scientific AI. Formal training is a plus, but not required.
• Ambitious, fast-learning, and excited to grow into a leadership role quickly.

You might be a fit if

• You were the person in school, research, or a previous job who made sure the group project actually shipped.
• You like being close to customers, product decisions, and execution details.
• You care about quality and get annoyed when important details are sloppy.
• You want a role where strong performance quickly turns into more responsibility.
• You’re excited by the idea of helping frontier AI labs build better systems for science.