Team Omnyra
Team Omnyra
Stanford University
Detecting the Next Generation of Synthetic Bioweapons
The Team
Emilin Mathew
Emilin (Emi) Mathew is a Stanford CS and Biology double major passionate about biosecurity and digital biology. She moves fast and thinks across disciplines, with experience as a SWE at Amazon, developer at SimVascular (open source cardiovascular tool used by 1K+ researchers), and designer of a text guided protein language model. She brings a product minded lens to frontier tech having consulted for OpenAI, DeepMind, and Google Data Commons. Emi works closely with the Stanford Biosecurity Program, leading events and organizing a 300+ person hackathon on emerging biological threats. She is excited to keep building toward a more secure and resilient bioeconomy.
Varsha Saravanan
Varsha is a Computer Science student at Stanford with a focus on Human-Computer Interaction, building high-impact, user-driven products at the intersection of technology, public health, and security. She brings product management experience from Rubrik, Develop for Good, and an incoming role at Microsoft, where she’s led cross-functional teams and launched solutions ranging from LLM-powered threat detection tools to EdTech and healthcare platforms. As a first author on GIS-based research advancing regional risk analysis and a co-developer of a digital health platform serving 500+ patients in California’s Central Valley, Varsha combines technical depth with real-world deployment. With a track record of building systems and two years of pro bono consulting for companies like Grammarly and Fizz, she’s now focused on transforming biosecurity infrastructure through intuitive, scalable, and defensible technology.
Overview
Federal Bureau of Investigation (FBI)
Problem Sponsor
Weapons of Mass Destruction Coordinators at the FBI who are working to prevent the illicit development of harmful pathogens cannot effectively monitor DNA synthesis to prevent harmful pathogens due to limited oversight and poor buyer tracking. This gap makes it harder to work with biotech companies and other partners to stop the misuse of synthetic genetics for biological threats
150+
Original Problem Statement
Number of Interviews
The Problem
What happens when synthetic biology meets malicious intent? That was the future-oriented challenge Team Omnyra tackled through Stanford’s Spring 2025 Hacking for Defense course.
Partnering with the FBI’s Weapons of Mass Destruction Directorate, students Emilin (Emi) Mathew and Varsha Saravanan on Team Omnyra addressed a rapidly evolving biosecurity threat: how to detect and mitigate the misuse of synthetic DNA in creating engineered bioweapons. Legacy screening systems struggle to detect cleverly altered DNA sequences used in synthetic orders, creating blind spots for law enforcement. Compounding the problem, the FBI’s reporting workflows for flagged threats were fragmented, unscalable, and reliant on outdated infrastructure.
“There isn’t a centralized system that looks at all the orders and routes them properly,” explained Saravanan. “We spoke to bioinformaticians, intel analysts, and field agents. Everyone knew the tools weren’t catching what they needed and reporting was a mess.”
The Innovation
Starting with 132 interviews during the course (and over 150 to date), the team’s Minimum Viable Product (MVP) evolved through several significant pivots. Originally focused on DNA screening, they expanded their scope to include a broader threat detection and reporting infrastructure that could support law enforcement, defense, and public health responses alike, such as AI-powered tools trained on synthetic pathogen datasets to flag engineered biological threats before they spread. These datasets don’t currently exist because the threats they aim to catch are speculative, emergent, and often deliberately engineered to evade known detection methods, making traditional data collection nearly impossible.
“What do you do when the threat you’re trying to stop doesn’t exist yet?” asked Mathew. “We had to simulate it.”
To bring their vision to life, Team Omnyra developed two breakthrough technologies. First, the team created a synthetic pathogen dataset, a novel framework for generating mutated, yet biologically plausible, sequences of known pathogens like Ebola or the 1918 Spanish Influenza virus. These sequences were designed to mimic how a malicious actor might engineer a virus to evade current tools, allowing the team to anticipate and simulate future threats in silico.
Second, they developed an AI-powered screening model trained on this synthetic dataset, capable of flagging DNA sequences that pose functional risks even when they don’t resemble any known threats. Their model was designed to work at scale, screening thousands of sequences efficiently, while detecting subtle, high-risk anomalies that traditional tools often miss.
To operationalize this capability for federal workflows, the team also built a reporting interface designed to streamline communication between flagged results and relevant FBI stakeholders, including WMD coordinators and bioinformatics analysts. This proof-of-concept interface is designed to enable secure, targeted reporting and to support high volume submissions, with the potential to process over 10,000 sequences in sub-minute response times.
The MVP received strong validation from FBI analysts and Defense Advanced Research Projects Agency (DARPA) program managers alike, with specific praise for its integration potential and forward-looking threat detection.
Team Omnyra’s Hacking for Defense Experience
As a team of undergraduates, Omnyra entered H4D with grit and curiosity.
“We’re computer scientists. We wanted to code on day one,” Mathew laughed. “But H4D taught us to sit in the problem first. We didn’t open an IDE until week six.”
Instead, they leaned into the course’s discovery methodology, interviewing over a hundred stakeholders and refining their assumptions weekly. The tough feedback loop, especially from their professor, Steve Blank, pushed them to grow.
“We got destroyed week one,” Mathew recalled. “But we took that seriously and never slipped again. By week six, we were asking to be in the main room every week to get the toughest feedback.”
Saravanan added, “Steve was like tough love, but so clear-eyed. He cut through the fluff and pushed us to find the right problem, not just a shiny solution.”
The team credits the teaching team and their FBI sponsor for giving them support and access that most undergrad teams rarely experience. “He met with us weekly, answered every email, and introduced us to more than 20 FBI experts,” said Mathew.
Presentation
What’s Next
Team Omnyra isn’t slowing down. They're currently part of the Defense Innovation Unit’s (DIU) Defense Innovation Summer Fellowship-Commercialization (DISF-C), where they continue to refine their prototype and explore go-to-market pathways.
“Our vision is a modular infrastructure for biosurveillance,” said Mathew. “From pandemic detection to battlefield biosensors, Omnyra aims to be a foundational layer for the biosecurity stack.”
They’re also co-authoring a Nature publication alongside over 20 leading researchers, including Princeton’s Director of AI for Accelerated Innovation and George Church, pushing forward what may become one of the most comprehensive frameworks for AI in biosecurity.
While the path ahead is long, and their field still nascent, Team Omnyra is in it for the long haul.
“When we get discouraged,” Mathew shared, “We ask ourselves: Do we believe this threat is real? Absolutely. So then: why not us? Why can’t we be the ones to fix it?”