[SCAN 2026 Final Interview] CertiScan: Tracking Virtual Assets with Practical Experience from CertiK

CertiScan Team. Source=CertiScan
The final of the world's first digital asset tracking competition, 'SCAN 2026', will be held on September 28 at 9 AM at Monaco Space in Seocho-gu, Seoul.

SCAN 2026 was established to enhance cybersecurity and investigative capabilities in digital assets. It is hosted by digital asset information company DAsset, with global blockchain data platform Chainalysis participating as a CTF (Capture The Flag) partner.

The final will feature the top 20 teams from the preliminary round (a total of 416 teams, 754 participants).

conducted written interviews with teams composed of global law enforcement agencies, security experts, and university students that advanced to the final.

*The teams advancing to the final are as follows. (In alphabetical order)

0xfun / 1nf1n1ty / 67 / BITSkrieg / CertiScan / DeFiHackLabs / EDCCS / Funds Went to Kerning City / hello / IBarelyKnowHer / J0y_B0y / lazarus2026 / Lv1x / Project Sekai / T34M1 / Tardigrades / Tornado Cats / youtiaos / Upside Academia / Chaos of KAOS

1. Please introduce the reason for forming the team and the background and expertise of the team members.

"The team is led by Yue Xue, the head of AI research at CertiK. Yue specializes in blockchain and smart contract security, particularly in applying large language models to program analysis, vulnerability detection, and formal verification.

They have developed several LLM-based security systems, including GPTScan, PropertyGPT, and Finite Monkey audit frameworks. They also have practical experience in smart contract auditing, static analysis, on-chain monitoring, anti-money laundering (AML) systems, and security incident investigation.

Bill Hsu is also part of CertiK, specializing in root cause analysis of blockchain attack transactions, on-chain fund tracking, and risk management. Based on real incident analysis experience, he helps the team quickly identify suspicious transaction patterns, reconstruct attack paths, and understand fund movements between various addresses and protocols.

Our expertise is highly complementary. Yue focuses primarily on AI-based analysis, security tools, and smart contract vulnerabilities, while Bill brings extensive experience in transaction-level investigation, incident response, and risk analysis."

2. You passed the preliminary round with 416 teams and advanced to the final. Please share your feelings about advancing to the final and the most memorable moment from the preliminary process.

"To be honest, most of us were excited. The preliminary problems were vast in scale and covered many categories, so the pace was very fast from the moment we started. Flags came out one after another, and we had to move on to the next problem without even having time to celebrate solving one. That immersive flow is what I remember most.

Advancing to the final among 416 teams is a great achievement, but what I really look forward to is the problems. I hope the final presents more difficult and creative challenges. That's when the conversation gets really interesting."

3. What was the most difficult problem or situation you faced while solving the SCAN 2026 preliminary problems? Please explain how you approached and solved the problem.

"The most challenging part of the preliminary round was not a specific problem but rather internal coordination within the team. The problem set was vast, and we were solving them quickly, but in the beginning, the distribution of tasks was honestly somewhat chaotic. There were overlaps in work, some problems were left unattended, and updates were scattered in the chat.

In a situation where the leaderboard was moving quickly, this friction actually led to significant time loss. The biggest lesson learned for the final is that we plan to operate much more systematically from the very start."

4. What tools, technologies, and analysis methods did you use for on-chain data analysis and fund flow tracking in the preliminary round? If you had any programs you developed or unique analysis methods, please share them.

"There was nothing particularly special. Our main strength was preparation. We built a kind of playbook in advance that informed how various types of on-chain investigations should be handled by AI agents. This was combined with general data sources, including helper scripts, RPC endpoints, block explorers, and various data query APIs.

The biggest advantage was that everything was already set up and tested before the competition started. This allowed us to focus on solving problems instead of wrestling with tools during the preliminary period. We also plan to open-source these skills so that others in the community can develop and improve this ecosystem."

5. Compared to other teams in the final, what do you think is your team's greatest strength or differentiated problem-solving approach?

"All teams in the final are strong, and there were several teams that were faster than us in the preliminary round, so it’s hard to say we have a special advantage. If I had to point out one thing, it would be that this work is closely related to our daily tasks.

Since our team is part of CertiK, a blockchain security company, investigating on-chain activities and tracking funds is something we do regularly in our auditing and incident response work. Many of the problems were similar to real cases we had encountered, which helped us quickly find our direction.

Additionally, CertiK has continuously invested in AI-based security research, and this mindset influenced the preparation process, including the aforementioned agent skills. I look forward to seeing how other finalist teams approach the problems and believe we can learn a lot in the final."

6. What preparations are you making ahead of the final on September 28?

"The main focus is on improving team collaboration. Based on the lessons learned from the preliminary round, we are defining clearer roles, establishing a more systematic task assignment process, and building a centralized channel to report and share progress and findings. The goal is to prevent overlapping work and ensure that all problems are immediately assigned to responsible parties.

On the technical side, we are continuously refining and testing AI agent skills, helper scripts, data query pipelines, and fund tracking workflows. We are also reviewing typical attack patterns and practicing with real incident cases under time constraints. We hope that by September 28, both the technical environment and collaboration processes will be perfectly prepared from the very start."

7. If you win first place in the final and receive a prize, what would you like to do with it?

"Part of the prize money will be used to support AI-based security research. This includes computing resources, model credits, and other infrastructure purchases needed to test and improve agents.

Some will be invested in new tools or open-source projects for on-chain investigation and blockchain security. Ideally, I hope this prize will not only recognize our achievements in the competition but also help develop tools that the broader security community can actually utilize based on what we learned. Of course, I plan to set aside some for the team members to celebrate together."