Job Posting Autopsy — Week 15: Bass Pro Shops AI Engineer II | Brian Curry

Series: Weekly Data Science Job Posting Review Company: Bass Pro Shops | Posted Title: AI Engineer II | Diagnosed Title: Staff AI Platform Engineer (with Data Engineering, MLOps, PromptOps, and AI Solutions responsibilities) — at mid-level pay, on-site only

Overview

Bass Pro Shops is hiring an “AI Engineer II” to its corporate IT team at its Springfield, Missouri headquarters. The posting requires 3–5 years of hands-on experience and a bachelor’s degree. No salary is disclosed. The role is fully on-site.

Bass Pro is privately held, owns Cabela’s (acquired in 2017), and operates one of the largest outdoor recreation retail footprints in the country. The “peak retail seasons” referenced in the posting are real — hunting and fishing seasons, holiday gift cycles, and Cabela’s-Bass Pro joint promotional periods all drive material spikes in inference and data load. So the operational scope is genuine. The question is whether it has been scoped to the right person.

The responsibilities section runs to over twenty distinct bullets covering AI engineering, data engineering, MLOps, LLMOps, prompt operations, agentic AI, AI governance, observability, and stakeholder consulting. There is also an on-call rotation. And the standard “all other duties as assigned” clause.

This is the most extreme example of title-scope compression the series has seen. T-Mobile’s Week 1 Senior Data Scientist was an ML engineer in disguise. Hallmark’s Week 5 Director was overstuffed but at least carried a director title to match the scope. Bass Pro’s posting is asking for a Staff-level AI Platform Engineer at the Engineer II level. Let’s break it down.

1. Role Identity Test

I mapped each stated responsibility to standard role archetypes: AI/ML Engineer, Data Engineer, MLOps/Platform Engineer, AI Architect, PromptOps Engineer, AI Governance Lead, AI Solutions Consultant, and SRE.

Collaborate with stakeholders to identify AI-driven automation opportunities and define success metrics → AI Solutions Consultant / Product. Discovery work, not engineering.
Translate business needs into technical requirements; design end-to-end AI workflows including data sourcing, orchestration, integration points → AI Architect. Architectural scope.
Assess data availability and quality across enterprise and third-party sources; design and validate pipelines with data contracts, lineage, and schema-drift detection → Data Engineer / Data Architect.
Perform data preparation and transformation in SQL or Python → Analytics Engineer / Data Engineer. Hands-on data work.
Design, develop, and deploy LLM-based, RAG, and generative AI solutions → AI/ML Engineer. Core function.
Maintain a versioned library of prompt/assistant patterns; help teams build and optimize assistants; ensure structured outputs, guardrails, prompt-injection resilience; run A/B evaluations for groundedness, accuracy, and safety with release gates → PromptOps Engineer. An emerging specialty in its own right.
Build and operate agents and semi-autonomous workflows with auditable actions, step limits, feature flags/kill switches, and human-in-the-loop approvals → AI Architect / Agentic Systems Engineer. Cutting-edge agentic AI work.
Implement model lifecycle management with experiment tracking, model registry, retraining, embedding/vector-index versioning, and rollback → MLOps Engineer. Full lifecycle ownership.
Integrate AI solutions via APIs and event-driven architectures; conduct load and cost tests; capacity planning and autoscaling for inference → Platform Engineer / Backend Engineer.
Containerization, orchestration, CI/CD, model lifecycle management → MLOps / Platform Engineer.
Implement automated testing (unit, integration, contract) for AI pipelines and services → Software Engineer / SDET.
Partner with Data Scientists to productionize models → ML Engineer. Standard handoff function.
Ensure responsible AI governance: RBAC/IAM, secrets management, PII minimization/redaction, payment-card and consumer-privacy regulations, data residency/retention, audit logging, bias detection, explainability → AI Governance Lead / Security Engineer.
Research and evaluate emerging AI technologies such as agentic AI, multimodal models, vector databases → Applied AI Researcher.
Implement observability (tracing, runtime telemetry, token/cost monitoring, alerting, dashboards), define and meet SLOs/SLAs and cost budgets, participate in an on-call rotation → SRE.
All other duties as assigned → Scope creep clause (familiar from Weeks 1 and 10).

The role spans eight distinct disciplines. Each has its own career ladder. Each commands its own market rate. The posting bundles all of them into a single mid-level IC.

2. The Title-Scope Mismatch

“Engineer II” is a mid-career rung. It implies someone who has shipped production work, can operate independently on scoped problems, and is being mentored by senior and staff engineers around them. It does not imply someone who:

Defines end-to-end architecture across multiple systems
Owns data contracts and schema-drift detection for enterprise pipelines
Operates a versioned prompt library and runs A/B safety evaluations as release gates
Architects agentic workflows with kill switches and human-in-the-loop approvals
Sets and meets SLOs/SLAs and cost budgets
Runs capacity planning and autoscaling for inference workloads
Makes recommendations that affect policies, procedures, and practices (the posting says this explicitly in the Independent Judgment section)

Every one of those is principal-engineer or staff-engineer scope. The “Independent Judgment” line is the tell — when a posting says the role makes recommendations affecting company policies, that is not Engineer II language. That is Staff or Principal language.

The 3–5 years experience requirement makes this worse. An engineer with 3 years of experience has not yet shipped enough production AI to have opinions on prompt-injection resilience patterns, vector index versioning strategy, or whether to use feature flags or kill switches for an autonomous workflow. They have not yet been on-call long enough to know what good observability looks like for an LLM system. The posting is asking for the output of a 7–10 year engineer at the price of a 3–5 year engineer.

3. The Skill Breadth Score

Counting distinct technical domains the posting requires hands-on competence in:

Python
SQL and data modeling
LLM/RAG system design
Generative AI frameworks
Vector databases and semantic retrieval
Major cloud platforms (AWS, Azure, GCP — the posting lists all three)
Snowflake / enterprise data warehousing
API development and integration
Event-driven architectures
Containerization (Docker)
Container orchestration (Kubernetes)
CI/CD pipelines
PromptOps tooling and patterns
Agent orchestration frameworks
MLOps tooling (experiment tracking, model registry)
LLMOps (prompt versioning, eval harnesses)
Observability stacks (tracing, telemetry, dashboards)
AI governance frameworks and compliance
Automated testing for AI pipelines (unit, integration, contract)

That is 19 distinct skill domains. With a target ceiling of about 8 for a mid-level IC, this scores 19/8 — roughly 240% over threshold. Hallmark’s Week 5 Director posting was 24/8. This posting comes close to that breadth at less than half the seniority and almost certainly less than half the pay.

Most striking: three different cloud platforms are listed as required familiarity. Pick a cloud. No engineer at any level operates fluently across AWS, Azure, and GCP simultaneously. Listing all three signals either an unsettled platform strategy or still determing the platform; although I have worked across all three.

4. PromptOps and Agentic AI Are Real — But This Is Not Where They Live

The posting deserves credit for naming PromptOps, agentic AI, vector index versioning, schema-drift detection, and observability for token/cost monitoring. These are genuinely current concerns. A year ago they would have been buzzwords; today they are real disciplines being built out at companies that are serious about production AI.

The data engineering scope also has a plausible business justification at Bass Pro specifically. The Cabela’s acquisition in 2017 produced one of the most complex retail data integration challenges of the past decade — two large outdoor retailers with overlapping but distinct customer bases, loyalty programs, inventory systems, and merchandising taxonomies. Even nine years in, that integration work is not done at most acquired-then-merged retailers. The “data contracts, lineage, and schema-drift detection” language in the posting reads like someone who has been burned by silent schema changes between legacy systems. Fair concern. Real work.

But none of this is Engineer II work.

PromptOps and agentic systems are not Engineer II disciplines. They are the work of staff engineers and emerging specialists. PromptOps in particular has no settled playbook — the patterns for “release gates for groundedness and safety” are being invented in real time at frontier labs and AI-native companies. Asking a mid-level engineer at a Midwest retailer to “contribute to and implement PromptOps” is asking them to invent the playbook, document it, and run it. That is a Staff Engineer with a focused scope — not an Engineer II with seventeen other responsibilities.

The same logic applies to the agentic AI bullet. “Build and operate agents and semi-autonomous workflows using established orchestration patterns, with auditable actions, step limits, feature flags/kill switches, and human-in-the-loop approvals for high-impact actions” is genuinely cutting-edge work. The orchestration patterns the posting references as “established” are largely two years old. Designing kill switches and HITL approval flows for autonomous systems is judgment-heavy work that requires having seen autonomous systems fail in production. A 3–5 year engineer has not had time to accumulate that experience.

5. The On-Site Requirement

The posting specifies on-site work at the Springfield, Missouri headquarters. No remote, no hybrid.

Springfield is not a large tech market. It has anchor employers in retail, healthcare, and financial services, but the local pool of engineers who have shipped production LLM/RAG systems with eval harnesses, vector indexes, agentic workflows, and observability is small. It is small in San Francisco. It is much smaller in Springfield.

The market data tells the story. Average AI/ML Engineer salary in Springfield, MO sits around $120,000 according to 2026 local data. The Springfield cost of living is roughly 30% below the national average. National mid-level AI engineer compensation runs $160,000 to $210,000 base, with median AI/ML engineering at $187,500. Adjusted for cost of living, Springfield’s market floor for a mid-level AI engineer with this scope sits closer to $130,000 to $150,000. With the LLM/MLOps/PromptOps premium, it should clear $150,000 to $170,000.

Whether the unposted compensation hits that floor is the question. Without disclosed pay, candidates have no way to evaluate whether the on-site requirement is being compensated. The on-site requirement alone removes roughly 40% of the senior AI candidate pool nationally, per recent staffing-firm data. Combined with no posted salary, the role will likely sit open longer than the company expects.

6. The On-Call Rotation Buried in the Last Bullet

The very last functional bullet of the responsibilities section reads: “participate in an on-call rotation as needed.”

This matters. On-call rotation is SRE work. Adding it to an Engineer II role bundles platform reliability ownership with feature engineering, architecture, prompt operations, governance, and stakeholder consulting. The candidate who signs on is committing to:

Designing and shipping AI systems
Defining the architecture for those systems
Operating the prompt library and eval harness for those systems
Owning the data pipelines feeding those systems
Being paged when those systems break

That is four jobs and a pager. At the Engineer II level, with no posted salary, on-site only, in a small tech market.

7. The Pay Transparency Pattern Continues

This is now the seventh Springfield-adjacent or Midwest posting in this series with no disclosed salary. Garmin, Hallmark, Burns & McDonnell, C2FO, the unnamed AI Solutions Architect, Koch — and now this one. The pattern is consistent enough to call out as a regional norm.

The cost of the missing range is asymmetric. The company saves nothing by hiding the number — internal HR has already set the band. Candidates lose the ability to self-screen and either over-apply (wasting the recruiter’s time) or under-apply (the company misses qualified people who assume the role is below their floor). The candidates the company most wants — engineers who already have multiple offers — are exactly the ones who skip postings with no posted range.

Missouri does not require salary disclosure. That is a legal answer, not a strategic one. Colorado, California, New York, Washington, and Illinois do require disclosure, and any company with employees in those states posting nationwide tends to disclose by default. The Midwest no-salary norm is a holdover from a tighter labor market that no longer exists for senior AI talent.

8. Compensation Assessment

Without a posted range, the analysis has to be reverse-constructed.

National benchmarks for mid-level AI engineers (3–5 years, LLM/RAG/MLOps experience): $160,000 to $210,000 base, with the LLM premium adding 10–15%. Median AI/ML engineering compensation across roles in 2026 is $187,500. MLOps engineers specifically average $165,000.

Springfield cost-of-living adjustment: roughly 30% below national. Applying that lever: $112,000 to $147,000 base for a generic mid-level AI engineer in Springfield. With the scope this posting describes — Staff-level architecture, PromptOps, agentic systems, on-call — the COL-adjusted band should be $150,000 to $185,000.

Local Springfield AI/ML engineer average: $120,000. That number reflects the existing local market, which is dominated by less specialized AI work than what this posting demands.

If the compensation hits the upper end of the COL-adjusted band, the role is compensated correctly for Springfield but the title is still wrong (it should be Senior or Staff, not II). If the compensation matches the local average of $120,000, the role is materially underpriced for the scope, on top of being mistitled.

One additional consideration specific to Bass Pro: as a privately held company, there is no public stock to use as compensation leverage. Public retailers like Target or Walmart can offer RSUs that materially expand total comp beyond base. Bass Pro cannot. That makes the base salary number — whatever it is — close to the entirety of the offer. With no equity to bridge the gap, the disclosed range matters more here than at a public competitor, and the decision to omit it does more damage to candidate self-screening.

9. What This Role Should Be

A more honest version of this posting would either split into multiple roles or restructure around a leadership hire who can own the function and build the team beneath them. Three reasonable rewrites:

Option A — Leadership hire, full scope: “Director of AI Engineering” or “Head of AI Platform” — 8–12 years, owns the full AI engineering function including LLM/RAG, MLOps, data infrastructure, and governance. Builds and manages a team of 3–4 engineers across the disciplines this posting bundles into one IC. This is the only single-hire solution that makes the scope honest. Posted band $175K–$220K.

Option B — One Senior IC role, focused scope: “Senior AI Engineer (LLM/RAG)” — 5–7 years, focused on LLM application development, RAG pipelines, prompt operations, and eval design. Data engineering and MLOps owned by adjacent roles or a separate hire. On-call optional or shared. Posted band $150K–$190K.

Option C — Two roles, two ladders: “Senior AI Engineer” focused on LLM/RAG application work and PromptOps, plus a separate “AI Platform Engineer” focused on data pipelines, MLOps tooling, observability, and on-call. Two roles, two budgets, two clear career paths.

Any of these rewrites would be more truthful, attract better candidates, and produce a healthier team structure. The current posting attracts either underqualified candidates who don’t realize what they’re signing up for, or overqualified candidates who use the offer as a baseline to negotiate elsewhere. The leadership option in particular reframes the problem entirely — instead of finding one engineer who can do everything, you find one leader who can build the team that does everything.

This is the most compressed posting the series has examined: Staff-level scope at Engineer II pricing, eight disciplines bundled into one IC, an on-call rotation buried in the last bullet, no posted salary, on-site only in a small tech market.

The genuine credit owed to this posting is that the AI work is real. PromptOps, agentic workflows with kill switches, vector index versioning, schema-drift detection, observability for token/cost — these are 2026-current concerns, not buzzword-bingo from 2023. Whoever wrote the responsibilities section understands what production AI actually requires.

The problem is they wrote it for one person at the wrong level. The hiring manager who needs everything in this posting needs a team of three to four engineers, with at least one staff engineer anchoring the architecture. Trying to compress that into a single Engineer II hire produces either a long time-to-fill or a hire who burns out fast.

For candidates: if you see this posting, the question to ask in the screen is “which of these eight responsibilities is the actual day-one priority, and which can wait six months?” If the recruiter cannot name a top three, the job has not been scoped. That is your signal.

For employers writing similar postings: split the role, post the range, and reconsider the on-site requirement. The AI engineering market in 2026 will not absorb compressed JDs at hidden pay grades.

Job Posting Autopsy is a weekly series that applies structured analysis to data science and AI job listings. The goal is not to mock employers but to help candidates make informed decisions — and to push the industry toward more honest role descriptions. Have a posting you’d like reviewed? Drop it in the comments.

Job Posting Autopsy — Week 15: Bass Pro Shops AI Engineer II