Hidden Job Families: The Roles Nobody Talks About

Project managers and programme managers together account for 4,229 job postings in our dataset. That's 37% of all tracked roles, more than ML engineering and data science combined. But delivery management rarely features in the conversations about "hot" tech roles, which tend to orbit data science, ML engineering, and product management.

The oversight extends beyond delivery. Analytics engineering, growth product management, research science, and platform product management are all substantial markets with distinct skill profiles, salary premiums, and hiring patterns. We mapped all 17 role types in our dataset to surface the ones that fly under the radar and explain why they matter.

The 17-role landscape

Our classification captures 17 distinct role types across three categories: data, product, and delivery. The volume distribution is uneven, and the tiers are instructive.

The top tier (2,000 or more postings each) includes core PM, ML engineer, programme manager, project manager, data engineer, and data scientist. These are the roles that dominate industry discourse and career advice.

The middle tier (400 to 1,100 postings) includes technical PM, AI/ML PM, data architect, platform PM, product analytics, and delivery manager. These roles are large enough to represent real career paths but small enough to be underrepresented in market commentary.

The specialist tier (under 450 postings) includes analytics engineer, growth PM, research scientist ML, and scrum master. These are niche roles with dedicated talent pools and, in several cases, meaningful salary premiums over their more common counterparts.

The skill overlap map

One of the most useful things this data reveals is which roles are genuinely distinct and which are variations on a theme.

Within the product category, the overlap is surprisingly high. Core PM and growth PM share 12 of their top 15 skills. Technical PM and AI/ML PM share 11 of 15. Technical PM and platform PM share 11 of 15. Growth PM and platform PM share 11 of 15. The product management variants are specialist applications of a common skill base, with domain expertise (AI, platform infrastructure, growth experimentation) layered on top.

Core PM and technical PM share 9 of 15 skills, which is the lowest overlap among product pairs. This is where the product category splits: core PMs are closer to the business side, while technical PMs are closer to engineering. But even here, the shared foundation is substantial.

Within the delivery category, the picture is similar. Programme managers and delivery managers share 11 of 15 skills. Project managers overlap with both at 10 of 15. The differentiation is scope and organisational altitude, not underlying capability. A project manager who learns to operate across multiple workstreams already has most of what a programme manager role requires.

Scrum masters are the most distinct delivery role, sharing only 6 to 8 skills with the other delivery types. The role's process orientation (Scrum, sprint management, continuous improvement) sets it apart from the broader coordination focus of programme and project management.

Within the data category, the story is more varied. Data engineers and data architects share 11 of 15 skills, making data architecture essentially an extension of data engineering into governance and strategic design. Data analysts and analytics engineers share 9 of 15, bridging through SQL, dbt, and visualisation tools. But data scientists overlap with analysts at only 5 of 15, and with analytics engineers at the same level. Data science is genuinely a different discipline, anchored in statistics and machine learning where the others are anchored in SQL and business intelligence.

Four roles worth a closer look

Research scientist ML: the most senior data role

Research scientist ML is the most seniority-skewed role in the dataset. 42% of postings target senior candidates, 26% target staff or principal level, and 9% target director and above. Only 15% of postings are junior or mid-level.

The role also has the strongest onsite bias among data roles at 25%, compared to 6 to 12% for other data subtypes. This likely reflects lab-based or research-centre hiring where physical presence is part of the work model.

US salary data puts the range at $187k to $254k, competitive with ML engineering ($185k to $265k) but with a higher floor. The skill profile centres on machine learning, Python, PyTorch, LLMs, deep learning, and reinforcement learning. This is where the most advanced ML work lives.

Data architect: governance meets infrastructure

Data architects account for 812 postings, making it the 10th largest role type. The seniority profile is striking: 41% senior, 26% staff/principal, 32% director and above. There is virtually no junior or mid-level hiring. If you're looking for this role, the market expects significant experience.

The skill profile blends infrastructure (AWS, Snowflake, Databricks, Azure) with governance (data governance and data architecture appear in the top 10). This is the role responsible for how data systems are designed and governed at an organisational level.

Growth PM: the most data-driven product role

Growth PMs represent 421 postings, 11% of all product roles. A/B testing is the top skill (appearing in half of all postings with skill data), followed by product strategy, SQL, and data analysis. Experimentation appears in the top 10, which doesn't happen for any other PM variant.

The salary range sits at $176k to $231k in US markets, slightly above core PM ($169k to $225k) despite being one-eighth the volume. The seniority profile is heavily senior at 70%, with minimal junior hiring at 1%.

Platform PM: where AI meets infrastructure

Platform PMs (671 postings) are the product role where AI has the strongest foothold. AI is the top skill with 78 mentions, followed by APIs at 69. The combination of AI awareness and infrastructure thinking (APIs, platform architecture) defines this variant.

US salary data shows $180k to $251k, a premium over core PM that reflects the technical depth required. The role has the most balanced seniority distribution among PM variants, with 55% senior, 19% staff, and 15% director-plus.

The title fragmentation problem

One reason certain roles stay hidden is that their job titles are wildly inconsistent. Core PM has 2,189 distinct title variations across 3,406 postings. The top title ("Product Manager") captures just 9.7% of all postings. ML engineer has 1,748 distinct titles, with "AI Engineer" (143 postings) and "Machine Learning Engineer" (137) splitting what is functionally the same role.

Data engineering is the exception. "Data Engineer" as a title captures 36% of all postings in that role type, followed by "Senior Data Engineer" at 19%. The role has the most standardised naming convention in the dataset, which likely reflects its longer tenure as an established discipline.

The fragmentation matters because it makes roles harder to search for. A candidate looking for growth PM opportunities needs to search for growth product manager, product manager (growth), senior PM growth, and dozens of other variations. The market is larger than any single search query suggests.

What falls outside the scope

Beyond the 17 tracked role types, our classifier categorises about 9,000 postings as out of scope. The largest segments within that bucket are software engineering at 14%, design at 6%, and product marketing at 8%.

Software engineers are correctly excluded as they're not specialised product, data, or delivery roles. But design leadership and product marketing represent meaningful career adjacencies. If the dataset's scope expands, these are the most likely additions.

The hidden roles aren't hidden because they're small. Programme management alone is 37% of the market. They're hidden because the career advice ecosystem orbits three or four role names, and the rest gets ignored. The data suggests the ignored half is where the opportunities are least competed for and the skill overlap with better-known roles is highest. The boundaries are softer than the titles suggest.

---

Based on 11,566 job postings from company career pages and aggregator sources, tracking 17 role types across data, product, and delivery families in London, New York, Denver, San Francisco, and Singapore. Skill overlap computed from top 15 skills per role type using company career page data only. Data covers November 2025 through February 2026. Full interactive dashboard at richjacobs.me/projects/hiring-market.