DataTalks.Club Podcast Wiki

Wiki

Career Transitions in Data

Archive-backed patterns for moving into data science, analytics engineering, data engineering, ML, AI engineering, and freelance data work.

Related Wiki Pages

Career Transition Job Search Career Growth Data Scientist Role Data Analyst Role Data Engineer Role Analytics Engineering Software Engineer to Machine Learning Academic Researcher to Data Science

Career transitions in data are moves from one working identity into another data role. The target can be analytics or data science. It can also be data engineering, machine learning, AI engineering, or freelance data work. In the DataTalks.Club archive, these transitions aren’t clean restarts. Guests keep parts of their previous work and turn them into evidence for the next role.

The repeated translations are practical. Project management becomes stakeholder and KPI work in Ksenia Legostay’s data-science transition. Marketing becomes funnel and BI knowledge in Nikola Maksimovic’s analytics-engineering transition. Software engineering becomes ML system building in Santiago Valdarrama’s software-to-ML episode.

QA becomes testing and project discipline in Alvaro Navas Peire’s QA transition. Academic research becomes statistics, domain data, and experimental reasoning in CJ Jenkins’ postdoc transition.

The archive’s common question isn’t “which course should I take?” It’s “what proof will make this transition believable for the target role?”

That proof can take several forms:

These examples recur in several interviews:

Core wiki routes:

Podcast starting points:

Common Definition

Across the archive, a data career transition means translating previous work into the language, artifacts, and responsibilities of a target data role. Ksenia Legostay’s route from project management into data science starts with customer-centric and KPI-driven work. It then moves through analytics, ML coursework, and CRISP-DM project framing.

Kaggle practice appears as another learning path, and production habits enter the story through Git and testing. Docker, deployment and clean code become part of the same transition (Project Manager to Data Scientist, 4:35-13:00 and 30:20-41:07).

That path connects the transition to data scientist work and job search because the candidate must show both analytical judgment and production awareness.

The same route appears in engineering-heavy moves. Santiago Valdarrama frames software-to-ML as adding machine learning to an existing engineering skillset, not discarding software engineering. His episode names coding as a core ML skill (Software Engineering to Machine Learning, 3:28-6:33).

He then pushes learners toward projects and data pipelines, with modeling and deployment coming next. Monitoring and APIs are deployment fundamentals, as are Docker and cloud providers (Software Engineering to Machine Learning, 17:25-49:23).

That’s why this page links software-to-ML transitions to MLOps, machine learning infrastructure, and Machine Learning Portfolio Projects.

The archive also treats transition evidence as role-specific. Slawomir Tulski’s data-engineering career episode ranks real work above side projects. Tutorial or certificate-only signals are weaker. His project advice favors a personal end-to-end data platform (Data Engineer Career in 2026, 42:08-57:35).

The project should ingest APIs or scraped data, then store and model the data before serving analysis (Data Engineer Career in 2026, 1:04:42).

Oleg Novikov and Nick Singh make the same hiring point from the interview side. CVs and take-homes should make contribution easy to evaluate. Project walkthroughs and behavioral stories should make role fit easy to evaluate. Case interviews test the same fit through scenarios (Data Science Interview Guide, 13:24-32:03, and Ace Data Interviews, 8:58-45:30).

Guest Differences

Guests differ on the best starting point. Some routes begin with formal study, while others begin with internal mobility or public projects. Community participation can also be the bridge. Freelance routes begin with market-facing client work. Ksenia Legostay includes formal data-analysis study and a part-time learning plan in her route (Project Manager to Data Scientist, 11:10 and 54:09).

Santiago Valdarrama argues for a problem-first route where software engineers start building before they feel mathematically complete (Software Engineering to Machine Learning, 8:12-29:05).

CJ Jenkins used structured learning and many CV iterations. Her decisive bridge was translating genomics, statistics, Bash, and R. Python, SQL, and messy research data became industry data-science evidence too (Postdoc to Data Science Lead, 1:28-17:14 and 40:02-43:44).

Guests also differ by target role. Nikola Maksimovic’s marketing-to-analytics engineering route stays close to SQL, BI, Looker and dbt. Data modeling and product analytics become part of the same analytics engineering path. A/B testing matters too (Marketing to Analytics Engineering, 7:18-33:46).

Alvaro Navas Peire’s QA route separates math-heavy ML from tooling-focused data engineering. He then uses cloud exercises and GitHub notes to make the transition visible. Technical projects and interview coaching help too (QA to ML and Data Engineering, 17:57-35:02 and 47:39-1:00:26).

Revathy Ramalingam’s AI-engineering restart shows a returner route. Current projects and community help update older software and telecom experience after a seven-year break. AI dev tools and take-home RAG-style assignments add current proof (Career Break to AI Engineer, 0:00-44:30).

Freelance transitions add a different disagreement. The proof isn’t only technical competence because the buyer also needs a reason to trust the person. Adrian Brudaru discusses early clients, pricing, scoping and intermediaries. Repeat business and reusable assets matter too (Freelance Data Engineering, 3:34-46:17).

Mikio Braun frames ML freelancing around network-driven leads and written proposals. Pricing tradeoffs and specialization define the work. Risk buffers and client outcomes matter too (Freelancing in Machine Learning, 7:53-52:45).

Dimitri Visnadi’s later data-freelancing episode puts market validation and rate evidence at the center. Recruiter channels and LinkedIn also define the path. Service positioning matters too (Data Freelancing Career Strategy, 10:50-1:01:02).

Transferable Skills Need Translation

The strongest transition stories keep useful prior skills but rename them for the target role. Ksenia Legostay turns planning and stakeholder communication into data-project framing. Business KPIs become decision-support evidence (Project Manager to Data Scientist, 22:32-32:43).

Nikola Maksimovic turns performance marketing into a BI and analytics-engineering path. Marketing already involved feedback loops and funnels. Dashboards and product questions were part of that work too (Marketing to Analytics Engineering, 2:53-14:14 and 38:27-41:50). These adjacent-business transitions are close to Data Analyst Role, Product Analytics, and Analytics Engineering.

For engineering transitions, Santiago Valdarrama says software engineers already have a hard ML skill because they can code and build systems. They still need data work, evaluation, ML tooling and deployment practice (Software Engineering to Machine Learning, 6:33-49:23).

Alvaro Navas Peire brings QA checklists and phone testing, while reporting and project discipline transfer too. He then adds ML and data engineering. Cloud familiarity and role-specific interview preparation matter as well (QA to ML and Data Engineering, 1:15-13:32 and 47:39-56:11). These routes connect to Software Engineer to Machine Learning, Data Engineering, and MLOps vs DevOps.

Academic transitions require the same translation but often start with stronger statistics and domain-data evidence. CJ Jenkins’ postdoc route includes population dynamics, GLMs, genomics files, and Bash. Data cleaning is part of the same research bridge. Her later gaps were deployment, APIs, Docker and Python production practice (Postdoc to Data Science Lead, 1:28-6:10).

Don’t treat publications as useless. Industry teams need to see the skills and artifacts behind the research. For the deeper route, read Academic Researcher to Data Science.

Portfolio Proof and Interview Story

Portfolio proof works when it’s specific to the role and easy to look at. Santiago Valdarrama tells software engineers to build and share real projects (Software Engineering to Machine Learning, 17:25-22:18).

Alvaro Navas Peire’s Zoomcamp projects and cloud exercises become job-search evidence. GitHub notes make the project narration easier to look at (QA to ML and Data Engineering, 24:57-35:02).

Revathy Ramalingam’s telecom network-slice capstone and GitHub work show current practice after a career break. Her AI-dev-tools prototype and PDF Q&A take-home add more current proof (Career Break to AI Engineer, 15:37-33:45).

Data-engineering portfolios need different evidence from ML portfolios, so Slawomir Tulski’s data-engineering episode favors end-to-end projects. A strong project includes ingestion and storage, plus modeling, serving and a useful personal or analytical consumer (Data Engineer Career in 2026, 57:35-1:04:42).

Merve Noyan’s Hugging Face episode shows an open-source portfolio path through contribution sprints, datasets, CI learning and Spaces. Community collaboration, PR workflows and public demos add more signals (Hugging Face Portfolio, 6:30-30:21 and 51:12). For candidates, these examples connect Data Engineering Portfolio Projects, Machine Learning Portfolio Projects, and Open Source Portfolio Evidence.

Interview story is part of the proof, not a separate soft layer. Oleg Novikov frames the CV as a landing page and asks candidates to show personal contribution. He treats take-home projects and behavioral stories as part of the same funnel. Case studies, SQL, coding and role targeting matter too (Data Science Interview Guide, 13:24-39:10).

Nick Singh adds that project walkthroughs should show ownership and impact. Business context and defensible technical claims come before case interviews and product-sense questions (Ace Data Interviews, 25:13-45:30). These episodes explain why the page treats transition as both a skills problem and a communication problem.

Internal Mobility and Community Routes

Some transitions happen inside an existing company before they appear on a resume. Nikola Maksimovic’s path through Ecosia starts with marketing work and Looker reporting. Conversations with the BI team, SQL learning, and BI projects come before the analytics-engineering title (Marketing to Analytics Engineering, 7:18-14:14).

Ksenia Legostay also recommends applying analysis at work and building a portfolio from real decisions when possible (Project Manager to Data Scientist, 32:43). These internal routes are useful when the current role already exposes the candidate to metrics and stakeholders. Data tools or product decisions make the bridge stronger.

Community routes create feedback, visibility, and referrals when the workplace doesn’t provide a clean bridge. CJ Jenkins names Berlin data-science meetups and community engagement as part of market entry (Postdoc to Data Science Lead, 33:48). Revathy Ramalingam uses DataTalks.Club community help and learning in public while rebuilding confidence after a career break (Career Break to AI Engineer, 11:00).

Dânia Meira’s community-building episode shows the broader mechanism. Networks and visibility help people see possible roles, while mentoring, panels and diverse talent pools help people step into leadership (Building ML Communities, 19:51-43:21). This connects transition work to Community Building, Career Growth, and Developer Relations.

Choosing the Target Role

The archive repeatedly warns that “data” is too broad for a transition plan. Rishabh Bhargava separates analytics from ML by goals and outputs. He also separates the roles by infrastructure and day-to-day work. Analysts build dashboards and reports. They also run ad hoc queries and make recommendations.

ML work moves toward models, APIs and predictions. It also adds SLAs, experiments and production feedback loops (Analytics to Production ML, 10:48-39:04).

Slawomir Tulski makes the data-engineering version explicit by separating platform-oriented and product-facing data engineering. He then emphasizes SQL, DevOps skills, cloud and processing engines. Cost awareness matters too, and he also warns against over-engineered platforms (Data Engineer Career in 2026, 11:54-42:08).

Target-role choice changes the learning plan:

Freelance and Consulting Transitions

Moving into freelance data work is a career transition from employee proof to buyer proof. Adrian Brudaru’s freelance data-engineering episode starts with leaving corporate and startup roles. It then covers the first client and income variability. Scoping work, pricing negotiation and repeat business come through networks (Freelance Data Engineering, 3:34-35:01).

Mikio Braun’s ML freelancing episode adds that proposals and qualification calls are part of the work. Scope alignment and risk buffers aren’t distractions from ML, and specialization and capacity management matter too (Freelancing in Machine Learning, 19:09-45:15).

The consulting route also changes what counts as a portfolio. Aleksander Kruszelnicki’s data-consulting story moves from product ideas to consulting after customer validation and user interviews. Market-size checks, network-first outreach, positioning and value-based pricing define the business (Data Consulting Business, 7:16-52:38).

Dimitri Visnadi’s data-freelancing strategy episode adds market validation and financial targets. Recruiter channels and LinkedIn help with acquisition. Rate research and subscription-style relationships matter. Notice-period planning matters too (Data Freelancing Career Strategy, 14:13-1:01:02). These episodes make Freelance part of career transitions rather than a separate business topic.

These pages connect the transition category to roles, hiring, and portfolio proof.