Full-Time
Forward Deployed Engineer
Updated on 9/4/2026
Goodfire
51-200 employees
AI interpretability tools and safety infrastructure
Compensation Overview
$200k - $325k/yr
San Francisco, CA, USA
In Person
In-person five days per week, with one company-wide remote week per month.
Bachelor's
 Software Engineering (1) |
|---|
Get referred to Goodfire
See people who can refer or advise you
- Strong software engineering skills with production experience in Python and modern machine learning frameworks, including PyTorch and JAX.
- Ability to work across research and engineering boundaries by understanding new techniques and shipping them.
- Ability to persist through ambiguity and learn quickly.
- Strong communication skills and the ability to explain technical tradeoffs clearly to engineers and non-technical stakeholders.
- Own technical delivery for partner engagements by building production-grade integrations of Goodfire's platform into customer environments, including API integrations, custom pipelines, internal tooling, and evaluation workflows.
- Design and execute creative, short-timeline pilots that demonstrate the platform's impact, often in parallel with or ahead of longer-term solution scoping.
- Serve as the day-to-day technical point of contact for partner engineering teams, work deeply in their codebases, understand their infrastructure, and build trust through technical competence.
- Bring field insights back to platform and research teams, identify repeatable patterns, and build shared tooling and playbooks that scale the Field Team.
- Take novel interpretability methods from the Foundational Team and make them work reliably in partner technology stacks by reading papers, prototyping techniques, and hardening them for production use.
- Technical degree or equivalent professional experience within artificial intelligence, machine learning, or a related field.
- Experience in a forward-deployed, solutions, or customer-facing engineering role; former technical founders are encouraged to apply.
- Track record of excellence in a high-growth startup or frontier artificial intelligence lab.
- Experience with large language models, including fine-tuning, evaluation frameworks, or agent development.
- Familiarity with interpretability, mechanistic interpretability, or model internals, including sparse autoencoders and feature steering.
Goodfire builds infrastructure and developer tools that allow users to understand, edit, and debug artificial intelligence models. These tools work by providing a practical interface for inspecting the internal logic of AI, enabling developers to identify and fix errors within complex systems at scale. Unlike many competitors that focus solely on theoretical research, Goodfire operates as a public benefit corporation that bridges the gap between science and practical application through specialized debugging software. The company's goal is to ensure the creation of safer and more reliable AI by making model behavior transparent and manageable for researchers and organizations.
Company Size
51-200
Company Stage
Series B
Total Funding
$207M
Headquarters
San Francisco, California
Founded
2024
Get referred to Goodfire
See people who can refer or advise you
Simplify's Take
What believers are saying
- Goodfire raised $150 million at a $1.25 billion valuation in February 2026.
- Silico launched April 2026 and opened broadly in August 2026, expanding market reach.
- Microsoft, Mayo Clinic, and Arc Institute validate demand beyond pure AI labs.
What critics are saying
- Anthropic, OpenAI, and Google DeepMind internal teams can copy interpretability features quickly.
- Silico’s long-horizon experiments are expensive, and bespoke pricing blocks broad self-serve adoption.
- If mechanistic interpretability disappoints, Goodfire becomes a costly research lab, not infrastructure.
What makes Goodfire unique
- Goodfire sells mechanistic interpretability, not generic AI tooling, through Silico and Ember.
- Silico, publicly available August 2026, lets users inspect and edit neurons during training.
- Goodfire already partners with Microsoft, Mayo Clinic, and Arc Institute on real deployments.
Help us improve and share your feedback! Did you find this helpful?
Benefits
Company Equity
Growth & Insights and Company News
6 month growth
↑ 0%1 year growth
↑ 8%2 year growth
↓ -3%San Francisco-based startup Goodfire has released Silico, a new tool that allows researchers to examine AI models and adjust their parameters during training. The tool uses mechanistic interpretability to map neurons and pathways inside models, enabling developers to reduce unwanted behaviours or steer outputs. Silico aims to transform AI model building from an opaque process into a more scientific, controllable one. By exposing internal mechanisms, Goodfire hopes to bring AI training closer to traditional software engineering practices, giving users unprecedented control over model development. The tool represents a significant step towards making large language models more transparent and debuggable, addressing longstanding concerns about the black-box nature of AI systems.
Goodfire has launched Silico, a mechanistic interpretability tool that allows developers to examine and adjust the inner workings of large language models. The tool enables users to zoom in on individual neurons or neuron groups within open-source models, run experiments to understand their function, and trace pathways to see how neurons interact. Using Silico, developers can modify parameters connected to specific neurons to boost or suppress certain behaviours. For instance, Goodfire researchers demonstrated how boosting neurons associated with transparency could change a model's ethical reasoning on disclosure decisions. The tool can also help filter training data to avoid setting unwanted parameter values. Goodfire aims to democratise techniques previously available only to top AI labs, making them accessible to smaller firms and research teams. Silico is available for a fee determined case-by-case.
Goodfire launches Silico: A game-changer for LLM debugging. Goodfire, a San Francisco-based startup, has unveiled Silico, a groundbreaking tool designed to enhance mechanistic interpretability in AI models. This innovative platform allows researchers and engineers to delve into the inner workings of large language models (LLMs), adjusting the parameters that define their behavior during training. According to Goodfire, Silico represents the first commercially available solution that facilitates debugging at every stage of AI development, from dataset creation to model training. CEO Eric Ho emphasizes the company's mission to transform AI model development from a mysterious process into a scientific discipline, addressing the existing knowledge gap between model deployment and understanding. Mechanistic interpretability is a cutting-edge approach that seeks to unveil the complexities of AI operations by mapping neural pathways and their interactions. This technique is gaining traction among industry leaders like Anthropic, OpenAI, and Google DeepMind, and has been recognized by MIT Technology Review as one of its Breakthrough Technologies. Goodfire aims not only to audit existing models but also to streamline the design process, eliminating the trial-and-error nature of model training. With Silico, developers can fine-tune LLM behaviors, such as reducing instances of hallucination, by exposing and manipulating the model's parameters. The tool employs automated agents to handle much of the interpretative work, making it accessible for users without extensive expertise. While Silico offers promising capabilities, experts like Leonard Bereska from the University of Amsterdam urge caution. He acknowledges the tool's utility but warns that the term 'engineering' might overstate its precision, suggesting it primarily enhances the existing alchemical nature of AI model training. Silico enables users to examine individual neurons within a trained model, allowing for targeted experiments and deeper understanding of how specific inputs affect outputs. For instance, Goodfire identified a neuron linked to ethical dilemmas within an open-source model, demonstrating how modifications can shift a model's responses. Furthermore, Silico can assist in steering the training process by filtering out undesirable influences from training data, ultimately helping to create more reliable AI systems. By democratizing access to advanced interpretability techniques, Goodfire aims to empower smaller firms and research teams to develop tailored models that meet their unique needs.
Goodfire AI, a US-based AI research lab, has raised $150 million in Series B funding at a $1.25 billion valuation. The company was founded in 2023 by Eric Ho, Dan Balsam and Tom McGrath. Goodfire describes itself as a research company using interpretability to understand, learn from and design AI systems. The startup's mission is to build the next generation of safe and powerful AI through understanding rather than scaling alone. The company focuses on making AI systems more transparent and controllable by examining how they function internally.
/PRNewswire/ -- Today, Goodfire—the AI research lab using interpretability to understand, learn from, and design models—announced a $150 million Series B...