Use cases in production make a data career
Holding the keys to the Golden Age of Data Careers (4)
I usually suggest avoiding data projects and proof-of-concept work. It’s much more valuable to join a team with multiple use cases in production and learn from that experience.
The Golden Age of Data Careers presents three significant opportunities for data professionals.
Impact: As the economy and society increasingly adopt data and innovative breakthroughs driven by Data & AI, the influence of data expertise and products grows, allowing you to determine the type and extent of the impact you want to create.
Advancement: As the data economy and the profession grow rapidly, you can advance your career, earn promotions, increase your income, and so forth.
Satisfaction: Growth and impact increase the likelihood of job satisfaction. It has been notoriously low in Europe for various reasons, but I expect that job satisfaction should significantly improve in the years to 2030.
Data professionals who understand this must, however, consider the importance of building a track record with real-world use cases in production. If you follow my assessment, this influences how you manage your career. In what follows, I explain how to hold the keys to the Golden Age.
What is a use case in production, and why is it important?
Who has use cases in production, and which type of company or organization has those with the most impact?
In what roles are you most likely to be successful, e.g., employee, freelance, consultant, founder?
Does the productionization of LLMs change the equation for data professionals and their roles, such as AI coding and agentic use cases?
In summary, what strategic choices do you need to be aware of as the most important?
Use cases in production
To recap what a use case in production is—you probably already know—it's a product or solution integrated into a business process or user experience, used continuously, and monitored and maintained over time. If the use case remains in production, it provides measurable value through revenue, cost savings, and operational efficiency.
Working on use cases in production gives you credibility and a track record - no matter what your exact role or contribution is. The best way to think about this is that if you have experienced several use cases in production, learned from the experience, gathered your insights, and moved on to a senior-level position. You are ready to start productionizing use cases, i.e., leading a major overhaul or improvement, and possibly even building from scratch.
At the start of the profession, from 2012 to 2022, most work focused on proof-of-concept and was often managed as a project. Correspondingly, there is this well-known anecdotal evidence that only a small percentage of the PoCs ever reached production. However, today you would want to work in a team that has a real-time feedback loop for both data and users.
Lack of deployment experience has often been seen as the main skills gap for data professionals, but there is no longer a good reason to accept that. By choosing an employer (company) that ships solutions, you take on a role in deploying, monitoring, and improving use cases in production. This approach helps you fast-track your promotion, transition into leadership roles, and significantly increase your value. Ideally, you should aim for a position that grants you ownership of at least one use case in production.
On the data production lines
Whatever your data role is, e.g., analytics, engineering, machine learning, or computer vision, being on the data production line means that you have a hand in one or more of the following.
Influence or own deployment. Deployment is where real risk resides and value is delivered. If you can’t deploy, you’re not part of the story. This involves being involved in MLOps tasks, such as deploying pipelines or monitoring, collaborating with infrastructure teams, or proposing production use cases.
Build infrastructure and ensure robustness. Early systems may work initially, but maintaining reliability over time is more challenging (e.g., drift, retraining, and upgrading), which helps build a reputation. You can use modular architecture, implement versioning, testing, monitoring, handle edge cases, and automate data pipelines.
Collaborate across functions. Business value is achieved when engineering, operations, and stakeholders are aligned. Learn the domain, engage with users, define metrics and SLAs, and communicate clearly with non-technical stakeholders.
Suppose you concentrate on teams that deliver use cases. This means that any role in startups, freelancing, or consulting is only valuable if you are also involved in turning those use cases into products or solutions.
Choosing roles and organizations
I have indicated that I believe that data careers are made on the production line, so to speak, and what does that mean for the other opportunities?
Going freelance?
I know senior data professionals who have launched freelance practices successfully in Europe. If you have a proven track record and deep industry expertise, you will find clients willing to pay up to € 2,000 per day. The key benefit I observed is that within 120 days, you can earn the equivalent of an annual salary and work remotely as much as you like. An additional attraction may be that you are not part of an organization or office politics.
Yet, only in the best scenario will you be contributing to productionizing use cases, and typically with ownership, but as a helping hand. Data remains a team sport, and so it is more likely that the internal team will own deployment and collaborate cross-functionally.
Starting as a consultant?
Consultancy firms have been acquiring data startups, developing algorithms, and establishing dedicated units for Data & AI consulting. Some consulting firms also have a core business that is data-driven in an obvious manner, e.g., the large accounting firms. Other firms have large knowledge databases that create new use case opportunities if LLMs are deployed. Yet, overall, the restrictions are similar to those of freelancing, and consultancy firms continue to operate in a project-based mode. I think this explains why large data product firms typically have little use for consultancies or freelancers.
In my mind, choosing a consultancy firm makes sense in two particular circumstances.
Launch your career, especially if you have been in academia long enough (e.g., Ph.D.), and a consultancy firm allows for an easier transition to industry. At the same time, you acquire expertise across domains and decide which industry you will join. Consultancies typically love smart PhDs.
Cash in as a Manager. Unfortunately, the “salary ceiling” in European industry is often still relatively low (never mind the astronomical sums paid in California). If you have the industry experience and contacts, a consultancy may value you at €150k or more. You also have the chance to make partner.
The LLM infrastructure shift
Gen AI is not just an add-on; it represents an infrastructure shift — a new form of data infrastructure. The most exciting part of this shift, in my opinion, is that anyone can now work with data because the basic skill needed is reading and writing to develop prompts and agents that make AI operational for small businesses, schools, and families.
In business, a one-person marketing agency can now deploy use cases in production based on the GenAI infrastructure. Additionally, freelancing gets a boost, not just with Agentic AI but also with the opportunity to use AI coding, meaning small teams and experienced individuals can make a significant impact. New use cases are emerging in categories such as product and process assistants or creative tasks. Sometimes these are described as shallow use cases but I expect that over time these use cases will be fine-tuned and monitored much better, and their value will increase.
If you are in a data-rich environment, the primary use of GenAI is accelerating production delivery, with the following notable shifts.
Faster deployments, such as LLM-based apps, can go live within days.
Use case agents, meaning the use cases, may be self-adapting, which can change how production is monitored and maintained.
Developing new skills like prompt engineering, vector databases, and retrieval-augmented generation.
Career strategy
Here is what I recommend as a focus for all those who have started a data career in 2022 or later.
Prioritize teams that deliver products and companies with multiple use cases in production. Any role in startups, freelancing, or consulting is only valuable if you participate in building, deploying, and monitoring data products.
Use your title and data role as a lever to expand your scope both technically (e.g., deployment and monitoring skills) and business-wise (e.g., domain expertise, bottom-line impact), and to accelerate your move to senior positions.
Aim to own at least one use case in production each year.
Good luck and enjoy the Golden Age!
AI Guild: Maximizing connections
You can use this companion to the Golden Age of Data Careers to enhance your connection with the data practitioner community. For example, consider joining the AI Guild or sharing your perspective.
Next Berlin AI Guild Dinner
The Future of Software Engineering. AI coding adoption. A special event with Deutsche Bank - Berlin Technology Centre on Wednesday, 08 October 2025.
