← MSBAi Home

MSBAi Curriculum Evaluation Report

Context

Three expert perspectives evaluated the MSBAi curriculum (36 credits, 15 months, 10 courses, Fall 2026 launch) against pedagogical soundness, industry hiring needs, and 2026 market trends. Each agent read all course syllabi, program documents, and conducted web research. This report synthesizes their findings into consensus recommendations.

What this report IS: A curriculum evaluation with specific change recommendations, grounded in learning science, hiring data, and competitive intelligence.

What this report is NOT: An implementation plan. The user will decide which recommendations to adopt before any files are changed.

Consensus Strengths (All 3 Perspectives Agree)

These need no changes — they are genuine competitive advantages:

Findings Ranked by Consensus Severity

TIER 1: CRITICAL (All 3 perspectives flag these)

C1. Quantum Course: KEEP (Decision Made) — Address Gaps Elsewhere

All three evaluators recommended replacing quantum, but decision: quantum stays (faculty commitment, forward-looking differentiator, unique in market). This means the agentic AI, RAG, and experimentation gaps flagged in C2 below must be addressed within existing courses rather than via credit reallocation.

Action items for quantum course itself (professor concerns):

• Ensure the Gies Coursera pre-reqs (stats courses) also cover the linear algebra basics needed for quantum
• Consider lightening the quantum prerequisites to “comfort with matrix operations” since students will have Python fluency from FIN 550
• The course remains a differentiator — no competitor offers this

C2. GenAI Course Needs RAG and Agentic AI Content (Within Existing 2 Credits) — RESOLVED

Professor: The Campus AI Framework lists “Agentic Systems & Workflows” as a track but no course delivers it. The strategy document promises “RAG skills” for every graduate but the GenAI syllabus doesn’t mention RAG.

Industry: “Everyone knows how to prompt ChatGPT. I need someone who can build an internal RAG tool.” LangChain is listed as “optional” — it should be required. The gap between “using AI” and “building with AI” is the hiring differentiator.

Market: Villanova already teaches multi-agent systems and “vibe coding” with Claude Code. By Fall 2026, a program branded “AI-first” that only teaches prompt engineering will face credibility questions. RAG is now as expected as knowing SQL.

Consensus recommendation (adapted for quantum staying): Restructure the GenAI 2-credit course to include RAG and agentic AI:

• Week 1: LLM fundamentals + prompt engineering (compressed from current 2 weeks)
• Week 2: RAG implementation (LangChain + vector DB) — make LangChain required, not optional
• Week 3: Agentic AI patterns (function calling, tool use, multi-agent intro) + AI governance basics
• Week 4: Capstone project (build an AI-augmented analytics workflow using RAG or agents) + ethics

Additionally, thread agentic AI concepts into BADM 576 (ML lifecycle week 7 could include LLMOps alongside MLOps) and the capstone (require AI tool building, not just AI tool usage).

C3. Statistics Foundation Before FIN 550: RESOLVED (Existing Coursera Pre-Reqs)

The gap: All evaluators flagged no formal statistics before FIN 550. Professor called it “like teaching calculus without algebra.”

Resolution: MSBAi will require two Gies College of Business Coursera courses: Exploring and Producing Data for Business Decision Making and Inferential and Predictive Statistics for Business. These are University of Illinois courses with proven content.

Action items:

• Document the Coursera stats pre-reqs in program/curriculum.md as a formal admission/pre-program requirement
• Add to courses/fin550_predictive_analytics.md prerequisites section
• Reference in program/design_principles.md under curricular constraints
• Ensure the admissions/onboarding materials communicate this clearly

C4. Add dbt and Modern Data Stack to BADM 558

Industry (CRITICAL): “dbt is the standard analytics engineering tool in 2026. Not teaching dbt is like teaching web development without CSS.” Snowflake has surpassed Redshift in new enterprise deployments.

Market (IMPORTANT): UT Dallas offers a Data Engineering track. The modern data stack (dbt + Snowflake/Databricks) is de facto standard at mid-size and large companies. The “Analytics Engineer” role is the fastest-growing analytics job title.

Professor: BADM 558 assumes Linux CLI skills not taught anywhere. Some content could be restructured.

Consensus recommendation:

• Replace or compress the EC2 deep-dive week with a dbt module (1 week)
• Introduce Snowflake as an alternative to Redshift in the data warehousing week (students see 2 platforms)
• Add a mandatory pre-course Linux CLI module (2-3 hours async) before BADM 558
• Consider renaming to “Cloud Data Engineering” to signal modern stack alignment

TIER 2: IMPORTANT (2 of 3 perspectives flag these)

I1. FIN 550 / BADM 576 Content Overlap — RESOLVED

Professor: Both courses cover regression, classification, decision trees, ensembles, feature engineering, clustering, time series, and deployment. FIN 550 tries to cover a full-year ML sequence in 8 weeks. The “spiral” is repetition, not deepening.

Industry: The deployment coverage is excellent (better than 90% of MSBA programs) but appears twice, in both FIN 550 and BADM 576.

Recommendation: Sharpen into ML I / ML II:

• FIN 550 (ML I): Supervised learning fundamentals only — regression, classification, evaluation metrics, simple feature engineering. Remove clustering, time series, and deployment.
• BADM 576 (ML II): Ensembles, regularization, unsupervised learning, NLP/text, time series, neural nets, deployment/MLOps. Absorb topics removed from FIN 550.

This reduces FIN 550’s cognitive overload and creates genuine spiral depth in BADM 576.

I2. Implement Oral Defense Before Launch — RESOLVED

Professor: “This is non-negotiable for academic integrity in an AI-enabled program.” The assessment strategy recommends 20-30% oral defense for projects and 40-50% for capstone, but course syllabi don’t include it. Without oral defense, project certification rests on artifacts that could be AI-generated.

Industry: Oral presentations verify understanding and build employer-valued communication skills.

Recommendation: Add oral defense (10-15 min presentation + 5 min Q&A) worth 20% of grade to at least 1 major project per course. Make capstone oral defense mandatory (30-40% of grade, panel format).

I3. Power BI Coverage Missing — RESOLVED

Industry: “In the Fortune 500 world, Power BI has overtaken Tableau in market share. Graduates who only know Tableau are missing half the market.” ~60% of enterprise analytics interviews involve Power BI.

Recommendation: Add Power BI as a companion tool in BADM 557 or BDI 513 — even a 2-week “same dashboard in Tableau vs Power BI” module.

Resolution: Power BI adopted program-wide, replacing Tableau. BADM 557 uses Power BI as primary BI tool.

I4. No Team Leadership Experience — DEFERRED

Industry: “The curriculum is almost entirely individual work. Analytics managers need to lead teams, manage timelines, navigate stakeholder disagreements.”

Recommendation: Make the capstone explicitly team-based (3-4 person teams) with a rotating project lead role and stakeholder management component.

Decision: Deferred. Team projects already exist in every 8-week course. Rotating lead role not a priority for Cohort 1.

I5. AI Governance Needs Dedicated Module — PARTIALLY RESOLVED

Market: Carnegie Mellon and Georgetown offer dedicated AI governance courses. IAPP’s AIGP certification signals governance as credential-worthy. Johns Hopkins has dedicated content. Ethics embedded across courses is good but different from governance as professional competency.

Professor: Ethics appears in final weeks of courses — signaling “afterthought.” Should be embedded in context, not standalone lectures.

Recommendation: If GenAI expands to 4 credits (C2), add 1 week of governance content (NIST AI RMF, EU AI Act, organizational AI oversight, model documentation). Require every capstone project to include an “AI Governance & Risk” section.

Partial resolution: Agentic AI course Week 3 covers NIST AI RMF + AI governance basics. GenAI stayed at 2cr so no room for a full dedicated module. Capstone requires AI governance section. Not a standalone course but governance is addressed.

I6. Elective 3: Superseded by General Elective (any iMBA) — RESOLVED

Update (Feb 2026): The dedicated “Elective 3” slot was replaced by a General Elective (any iMBA course) in Spring 2027, Weeks 9-16. Students select from the broader iMBA catalog. MSBAi-specific electives (Data Engineering, MLOps, etc.) remain under consideration for future cohorts — see Future Electives.

Original finding: “Data Engineering fills both critical gaps: experimentation/MLOps and modern data stack.” Option B (MLOps + A/B testing) is the industry veteran’s first choice; Option A (Data Engineering with dbt) is the market analyst’s recommendation.

Market: Analytics Engineer is the fastest-growing job title. dbt + Airflow + Kafka directly maps to job postings.

Original recommendation: Confirm Option A (Data Engineering) for Cohort 1. If two electives become possible in Year 2, add Option B (MLOps + A/B testing). This combination fills both critical gaps.

I7. BADM 557 Placement in Summer 2027 — CLOSED (Keeping Summer)

Professor: As the sole summer course, BADM 557 runs in isolation. Its business communication and BI skills would improve student work in every subsequent course if taught earlier. However, it genuinely needs FIN 550 as a prerequisite for its classification/clustering components.

Recommendation: Evaluate whether BADM 557 can drop the ML components (classification/clustering — these are also in FIN 550) and move to Spring 2027 weeks 9-12, teaching BI/dashboards/executive communication earlier. If ML components are essential, keep current placement but fill the empty Spring 2027 weeks 13-16 with a professional development module (AWS certification prep, career branding workshop, or capstone scoping).

Decision: Keep BADM 557 in Summer 2027. FIN 550 prerequisite is essential.

I8. Operationalize AIAS Levels in Course Syllabi — RESOLVED

Professor: The AIAS framework (AI usage levels 0-4 per assignment) exists in the assessment strategy but no course syllabus specifies levels for individual assignments.

Recommendation: Annotate every assessment component with its AIAS level before Fall 2026 launch. Baseline: quizzes = Level 0, Project 1 = Level 1-2, later projects = Level 2-3, GenAI course = Level 4.

Resolution: AIAS levels added to all 9 course syllabi (BADM 554, BDI 513, FIN 550, BADM 557, BADM 558, BADM 576, Agentic AI, Quantum, Capstone). Each assessment component specifies its permitted AI usage level. Source citation (Perkins et al., 2024) added to Assessment Strategy. AIAS requirement added as Constraint 10 in Design Principles.

TIER 3: MINOR (Noted but lower priority)

The Highest-Impact Changes (Given Decisions)

With quantum staying and stats pre-reqs resolved, the top priorities shift to:

1. Restructure GenAI 2-credit course to include RAG + agentic AI (C2) — ✅ Done. Renamed to Agentic AI for Analytics; RAG, agentic patterns, LangChain, governance all included.
2. Add dbt + Snowflake to BADM 558 (C4) — ✅ Done.
3. Sharpen FIN 550 / BADM 576 boundary (I1) — ✅ Done. FIN 550 = ML I (supervised only), BADM 576 = ML II (advanced + LLMOps).
4. Document Coursera stats pre-reqs (C3) — ✅ Done. Documented in CURRICULUM.md and FIN 550 prerequisites.

Implementation: What Changes in the Files

This is a decision document, not an implementation plan. Once you decide which recommendations to adopt, the changes would touch:

Verification

After any changes:

1. ✅ Total credits still sum to 36 (no credit changes — quantum stays, Agentic AI stays at 2cr)
2. ✅ Prerequisite chain remains valid (Coursera stats pre-reqs documented)
3. ✅ No semester exceeds 12 credits
4. ✅ L-C-E progression still holds at program level
5. ✅ AI_FIRST_STRATEGY.md promises (RAG, agentic systems) match actual Agentic AI course content
6. ✅ FIN 550 and BADM 576 have clean topic boundary with no overlap
7. ✅ BADM 558 dbt/Snowflake additions reflected in CURRICULUM.md skill progression
8. ✅ Coursera stats pre-reqs listed in CURRICULUM.md and FIN 550 prerequisites

Sources (from market research agent)

• Gartner: 40% of enterprise apps with AI agents by 2026
• PwC: AI-skilled workers earn up to 56% more
• Poets&Quants: Villanova MSBAi (agents, RAG, vibe coding)
• UC San Diego Rady: LLMs in every facet
• USC Bovard MSAA: AI at its core
• Carnegie Mellon: Responsible AI & Governance course
• IAPP AIGP 2026 update
• IMF: 1.6M unfilled AI positions globally