There are different types of data scientists, so agencies need to be clear on what their needs.
Ann Irvine is principal data scientist at RedOwl.
The FBI announced it's hiring its first senior-level data scientist, meaning the agency is officially recognizing that insight into data is a critically needed piece of its risk management strategy.
However, as someone who has worked in the field for years, specializing in natural language processing and machine translation, I understand all too well the work the bureau has cut out for it in finding the right candidate and integrating that person into its organization—challenges that might prove larger than the agency originally anticipated.
Data Scientists Aren't Created Equal
First and foremost, data scientists across organizations and industries have wildly different backgrounds, skillsets and job responsibilities. That’s why it’s crucial for the FBIーor any organization in search of this type of new hireーto be very specific about what its data science needs are, what the new hire’s responsibilities will be, and how that person will fit into the organization’s day-to-day workflow.
For example, on my team at RedOwl, we have three types of data scientists who support our technology’s capabilities to prevent insider threats such as rogue trading on Wall Street or IP leaks in the enterprise. Our field data scientists work closely with customers to identify their needs. Our product development data scientists contribute directly to the core software. And our product strategy data scientists define the product’s analytic road map.
While each of these teams must have knowledge of mathematics and machine learning, subject-matter expertise in cybersecurity, strong communication skills and experience in software development, the importance of each varies based on the specific job role.
For the FBI’s new hire, this is just one reason why it’s critical for the agency to be clear about its specific needs. Doing so will allow it to better identify candidates who will be a good fit among what I’m sure will be a very large and diverse pool of applicants.
Communication is Key
Seeking a senior-level data scientist, the FBI should also ensure it finds a candidate with strong communications skills, as this person will advise and represent the agency on committees throughout the intelligence community. In the enterprise world, senior-level data scientists may similarly be held accountable for reporting their team’s findings to the C-suite and the board of advisers.
In this capacity, the data scientist will also need to serve as a reality-checker for what is and what’s not possible to do with data science: Could this type of data answer that type of question? How much data would be needed? What types of algorithms should we try? How will we evaluate them? How much work would it take to prototype an implementation? What kind of mistakes will the algorithm make? Has this or a similar problem been solved before?
A senior data scientist must also have good instincts about what challenges are achievable with different levels of effort and, for the FBI, will need to be able to explain their reasoning clearly throughout the intelligence community.
The FBI’s job description doesn’t mention whether the data scientist will be expected to implement technical solutions and, if so, whether software would be productized for ongoing use or be prototyped for individual projects.
However, the answers to these questions have a substantial impact on the skillset that the FBI, or data scientist-seeking company, should target. Regardless, the best candidate will be one who is familiar with the details of possible technical approaches, even if that person will not be in a position to implement them by him or herself. This includes familiarity with open source machine learning and data visualization tools.
To avoid the common mistakes I’ve seen many organizations make when they hire their first data scientists, the FBI should first deliberate on how it integrates the data scientist into the organization. It will be critical for the new hire to have easy access to relevant data sources so they aren’t bogged down with locating, cleaning and organizing data.
Additionally, it may be tempting for the FBI to quickly engage the new hire in many projects—both within the bureau and throughout the intelligence community. However, my advice to them and all other organizations making this type of new hire is to make sure there’s dedicated time built into the onboarding process to review prior work, get to know existing data sets, stay up to date on technologies, and implement technical solutions.
The introduction of data science into the FBI is exciting and will undoubtedly make a big impact on the agency’s work. With thoughtful planning, interviewing and execution, the FBI’s newest senior-level employee will be able to be effective right away.
NEXT STORY: What's In Pentagon Newly Released IT Vision?