Despite the continuous evolution of technology, there’s been virtually no decline in the number of fatal workplace injuries in the United States for the past 21 years.
Data from the U.S. Bureau of Labor Statistics reveals a disturbing pattern: the more serious a category of workplace injuries is, the less progress has been made over the past two decades to reduce related incidents.
“This persistence highlights a fundamental problem — identifying and mitigating serious injury and fatality risks at scale remains too difficult to do manually,” Dr. Julia Penfield, vice president of research and machine learning at VelocityEHS, explained.
But with AI and machine learning, that might be about to change.
Penfield, whose previous work at an electrical power plant involved personally witnessing devastating injuries, believes that these technologies offer a generational opportunity for VelocityEHS to change the status quo by offering customers an automated, efficient approach to risk assessment.
About VelocityEHS
VelocityEHS’ software is designed to help organizations reach their environmental, health and safety goals, offering AI-driven capabilities related to chemical inventory management, contractor safety and permitting, environmental compliance and more.
“We’ve developed a portfolio of highly customized AI solutions that consistently outperform generic GenAI models in our domain,” Dr. Penfield said. “That includes language models tuned for safety narratives, vision systems that detect environmental hazards, and hybrid models that merge structured and unstructured inputs.”
For Dr. Penfield, the opportunity to close what she describes as an “unacceptable gap in safety performance” is what first drew her to her employer, where she merges her passion for environmental, health and safety field realities with deep AI expertise — but what keeps her there is her team, which is committed to continuous learning and technical excellence made possible by her employer’s investment in hands-on learning, experimentation, and encouragement to test out emerging techniques in practical contexts.
“There’s a real sense of purpose — this isn’t just model accuracy, it’s about saving lives and enabling safer work,” Dr. Penfield said.
The Impact of AI Innovation at VelocityEHS
VelocityEHS’ technologists don’t build AI solutions purely for the sake of innovation.
“Our team is deeply mission-driven,” Dr. Penfield said. “We’re here because we believe in preventing serious injuries, fatalities and long-term harm.”
In team meetings, Dr. Penfield said they often revisit the impact of their work: reducing fatalities, amputations, temporary and permanent disabilities, and even the less severe but still deeply disruptive injuries.
“We’re also motivated by the challenge of doing applied AI in one of the most human-centered and safety-critical domains,” Dr. Penfield explained.
So it’s no surprise that the company has attracted a group of lifelong learners, including many — like Dr. Penfield — who have transitioned from other industries or academia in order to thrive in a space that is technically rich and socially meaningful.
With the help of AI and ML, VelocityEHS offers organizations a “virtual army” of EHS professionals to identify and assess risks early. It also provides frontline workers with a smart assistant to better prepare for tasks and gives incident investigators a digital colleague to post-process events more deeply and efficiently. These capabilities, among others, fuel the company’s efforts to radically redefine how workplace injury risks are predicted, handled and processed.
“Our philosophy centers around driving value to customers, building trust, and ensuring transparency,” Dr. Penfield said. “Every model or tool we build is evaluated not just for accuracy, but for whether it helps customers make better safety decisions and enhances their experience.”
To that end, Penfield said her team focuses on customized models rather than generic solutions, ensuring that their systems reflect the realities of EHS operations. Before deploying anything at scale, they conduct feasibility studies and structured pilots to validate utility and safety.
“We also take data ethics seriously,” Penfield said. “We have carefully developed internal processes and tooling — in collaboration with our compliance officers — to ensure that customer data is never used for model development unless they explicitly volunteer to participate in that process. This builds trust and sets a high bar for responsible AI.”
According to Dr. Penfield, the company recently hit a major milestone in its AI journey with the launch of its Potential Serious Injury and Fatality prediction tool, which leverages natural language processing and structured risk modeling to identify incidents that could have resulted in severe outcomes, even if the actual outcome was less serious. The solution, which required ongoing collaboration between data scientists, safety experts, and software engineers, was designed to empower customers to surface hidden risks and intervene earlier.
“It’s also a prime example of how our AI roadmap aligns with a broader strategy of proactive safety enablement,” Dr. Penfield said. “Without AI, customers would still be relying on manual review of vast amounts of reports, audits, and incident data — slow and inconsistent work that can miss critical signals.”
AI changes the status quo by introducing predictive models and leading indicators that help customers act before incidents occur, Dr. Penfield explained.
“We support them by recommending corrective actions, identifying patterns across locations, and highlighting underreported risks,” she said.
This level of decision support was not feasible at scale before AI, which helps users move from compliance-driven operations to continuous safety improvement, driven by data and supported by intelligence.
When AI isn’t being used to enhance the company’s product capabilities, it’s helping the company’s technologists work internally with greater speed and efficiency. For example, the organization uses internal AI tools to streamline data preprocessing, model validation, document classification and safety taxonomy curation, significantly reducing the cycle time for developing and testing new AI features across its product lines.
“By deploying these intelligent accelerators, we multiply the impact of our research and deliver innovation faster and more reliably,” Dr. Penfield said. “AI is not just the output — it is increasingly the infrastructure that supports our product development machine.”
“AI is not just the output — it is increasingly the infrastructure that supports our product development machine.”
At VelocityEHS, AI is not only used to enhance product capabilities — it is also a key enabler of accelerated AI development itself, Dr Penfield added.
“We are building AI to help us build better AI,” Dr. Penfield said.
She’s especially excited to see how AI is transforming the way different departments work. To support the company’s sales, marketing and customer support teams, Dr. Penfield’s team has been building tools like smart proposal generators and document summarizers, signaling what she considers a “paradigm shift” that stretches beyond production innovation and into smarter operations across the board.
“In addition to pushing state-of-the-art models into our products, my team now operates as a mini R&D group that boosts the entire company’s productivity,” Dr. Penfield said. “The value goes beyond product innovation — it’s about enabling smarter operations across the board.”
According to Dr. Penfield, this internal transformation underscores how AI can be a horizontal capability, not just a feature.
“We’re just scratching the surface of what’s possible,” she said.
The Team Culture at VelocityEHS
Inspired by the potential to make an impact on others, the team at VelocityEHS transforms their own lives in the process.
Working at the intersection of AI and EHS, the company’s technologists engage with professional AI communities and attend top-tier conferences. They also frequently contribute to the world of academia and, on occasion, get to see their work emblazoned with a trademark symbol.
“Because we’re a pioneer company in applying AI to the EHS space, our team members often have the opportunity to publish in prominent peer-reviewed journals — a rarity for professionals outside academia,” Dr. Penfield said.
The company has been awarded 12 patents by the U.S. Patent and Trademark Office for its use of AI and ML to help prevent workplace injuries and accidents. Additionally, the organization recently published two scientific papers in the journal Nature, both of which list Dr. Penfield as an author.
And because her team works closely with compliance officers, ergonomists, and safety professionals, the team is exposed to deep subject matter expertise that shapes their thinking and sharpens their impact, Dr. Penfield said.
“They also get to own projects end-to-end, from problem definition to deployment and user testing,” Dr. Penfield said. “That level of autonomy is hard to find elsewhere.”
To stay up to date on current trends or developments in their field, Dr. Penfield said her team engages with professional AI communities, academic research, peer-reviewed publications, and attends top-tier conferences like NeurIPS, ICML and ACL. They also have internal reading groups that dissect the latest technical innovations and evaluate how they apply to their domain.
“Importantly, we make a deliberate effort to stay away from pop-AI sources, such as podcasts or over-hyped platforms,” Dr. Penfield said. “Instead, we emphasize rigor and depth, staying grounded in professional and research-oriented spaces that align with our mission.”
“We emphasize rigor and depth, staying grounded in professional and research-oriented spaces that align with our mission.”
The company’s technical architecture, which blends domain-informed rules with learned patterns, makes team members’ work doubly fulfilling, ensuring that both performance and transparency remain at the forefront of the company’s tech initiatives.
“Think of it like a safety GPS: the ML components assess traffic (risks), while expert systems ensure the route complies with the rules of the road,” Dr. Penfield explained.
While VelocityEHS’ technologists undoubtedly work together to bring the company’s cutting-edge solutions to life, they also get to lean on their own individual talent and expertise to drive change. Whether they’re handling problem definition or user testing, everyone who joins one of the company’s tech teams gets to add their own strokes of genius to the bigger picture — and own their impact in the process.
“This tight integration between domain expertise and technical rigor is something we’re extremely proud of — and it’s what makes our solutions reliable in high-stakes environments,” Dr. Penfield said.
