Resume Teardown #78: CS-Adjacent Undergrad, 78%, Real PM Craft Missing One Layer
This is part of our Resume Teardown series where we score real PM resumes (anonymized) and break down what the evaluation found.
TL;DR: A chemical engineering undergrad at a top technical institute, targeting PM and AI PM roles, scored 78%, which is high for a student. The product craft is genuinely present: discovery, requirements, prototypes, roadmaps. The one missing layer is what happened after anything was built.
The Resume
Background: Third-year chemical engineering student at a top technical institute, pivoting hard toward product. Three PM-adjacent internships: product discovery and an MVP at a campus alumni relations office, strategy and competitor analysis at a solar energy company, and AI product design at an enterprise AI company. A deep bench of product and data projects, multiple national case-competition placements, and sustained campus leadership with real budgets and teams.
What looked good on the surface:
- Genuine PM craft shown through work, not tool lists: discovery, user segments, requirements, information architecture, wireframes, Figma prototypes, roadmaps
- Unusually concrete AI product work for an undergrad: an offline retrieval-augmented generation (RAG) assistant with retrieval accuracy, latency targets, privacy constraints, and defined AI product KPIs
- National case-competition results, including a first-place finish, showing structured problem framing under pressure
- Substantial campus leadership: large teams, real budgets, external partnerships, and end-to-end event delivery
Score: 78%
Why This Student Scored Higher Than Most
Most student PM resumes we tear down score in the 50s and 60s because they describe engineering or coursework and leave the reader to infer product thinking. This one scored 78% because the product thinking is on the page, in the candidate's own bullets.
The skills dimension scored 80%, which is rare at the undergrad level, and it is earned. The resume shows discovery, requirements, prototyping, and workflow design as activities this candidate actually did, not as skills in a list. That is the single biggest differentiator between this resume and a typical student resume.
So the interesting question is not "why is the score low." It is "what is the one thing keeping an interested hiring manager from a clear yes." The verdict names it directly:
The hands-on product evidence is stronger than usual at this stage, but I cannot yet tell which work shipped or changed a user or business outcome.
That is a single, specific, fixable gap. Everything below is about closing it.
The Core Gap: Built, But Then What?
The strongest internship and project bullets describe what was designed or built, and stop there. They rarely show what happened once a user or stakeholder touched it.
"Built AWS-powered QR automated MVP, digitizing manual verification through a scalable event access system"
This is a real MVP with a real purpose. But the bullet ends at "built." Was it used at an event? How many entries did it process? What was the manual process before, and how much faster or more reliable did it get? The outcome almost certainly exists, because this was built for a live campus context.
"Owned product design for offline RAG assistant, achieved 95% retrieval accuracy via SQL and Ollama architecture"
The 95% retrieval accuracy is a strong metric. But the bullet does not say who the assistant was for, what retrieval task they had, how quality was evaluated, or whether it was prototyped, tested with users, or deployed. The difference between "I designed a RAG assistant" and "I shipped a RAG assistant that N users relied on for X" is the difference between a project and product ownership.
The fix pattern:
Before: "Built AWS-powered QR automated MVP, digitizing manual verification through a scalable event access system"
After: "Shipped a QR-based access MVP that replaced manual entry verification at [event], processing [N] check-ins and cutting average entry time from [X] to [Y]."
Same work. The after version shows the thing was real, used, and better than what it replaced. That is the layer the hiring manager is asking for.
The AI Work: One Guardrail Decision Away From Excellent
Because this resume triggered the AI PM signal, the RAG assistant is the candidate's strongest lever, and it is close to excellent.
"Defined AI product KPIs and SLAs, targeting <3s latency across retrieval quality, faithfulness and query success"
"Designed privacy-first RAG product with 0 external APIs, defining user flows, requirements, and retrieval logic"
This is genuinely good AI PM thinking for an undergrad: KPIs, latency targets, faithfulness as a metric, a privacy-first constraint. What is missing is the single most important AI product decision: what happens when the model is wrong.
The resume shows retrieval performance and latency, but not the failure-mode design. For a retrieval assistant, that means: confidence thresholds, source citations, an escalation path, or a human-review fallback for low-confidence answers. AI PM interviewers probe exactly this. Adding the guardrail you designed for unreliable answers would turn strong AI work into interview-ready AI product work.
The Data Project That Reads as Data Science
One project is strong technically but framed away from PM:
"Built segmentation-to-recommendation pipeline converting 1M+ transactions to next-best-product offer for users"
"Deployed XGBoost classifier achieving 95% accuracy and 96% macro-F1 with comprehensive bias auditing"
The modeling rigor is real, and the bias auditing is a nice signal. But as written, this reads as a data science project, not PM evidence. For a PM resume, lead with the user and business decision, not the model metrics.
What was the business problem? Who acts on a next-best-product offer, and how? Did the recommendation change what a business team would do? The macro-F1 score matters to a data scientist. A PM hiring manager wants to know what decision the pipeline enabled and what it would change in production.
The Missing Summary
There is no summary, and for a candidate with this much range, that is a real cost.
The resume spans chemical engineering, AI product work, solar strategy, data projects, and campus leadership, with no line connecting them into a product direction. A recruiter has to assemble the story themselves, and the most common conclusion is "talented, but unfocused."
Two lines before the education section would fix this. Position explicitly: an undergrad at a top technical institute building data- and AI-enabled products through user research, workflow design, and prototyping, targeting APM or AI PM roles. That one addition reframes the breadth from scattered to intentional.
Dimension Scores
Skills and Tools: 80%
The highest dimension and the strongest asset. Real PM craft shown through work: discovery, requirements, information architecture, wireframes, prototypes, roadmaps, and workflow design. The AI product work is unusually concrete for a student. The one gap is the AI guardrail decision described above.
Leadership and Impact: 76%
Credible ownership for an undergrad: product discovery, user journeys, roadmap definition, and a live MVP. Sustained campus leadership with large teams and real budgets. The gap is the same throughline: bullets describe what was designed or built, rarely what happened after users or stakeholders used it. Add pilot usage, stakeholder validation, or delivery status.
Experience and Background: 74%
A strong student mix of PM-adjacent internships, technical product work, case competitions, and leadership. The candidate has clearly sought product responsibility rather than coasting on the engineering degree. The gap is the missing one-line narrative connecting it all into a focused direction.
Domain Expertise: 71%
Broad rather than deep. Signals span enterprise AI, climate and energy operations, financial services, procurement, and agriculture. For a student this breadth is fine, but a recruiter cannot tell which PM openings the candidate is most prepared for. Pick one primary area (AI-enabled enterprise workflows is the strongest candidate here) and group supporting evidence around it.
ATS Readiness: 65%
Pass: Standard section headers, no material spelling errors.
Warning (acronyms): Several role-specific and regional acronyms are expanded only in endnotes. Terms like the alumni-relations office abbreviation, the geospatial assessment tool, and the chess-rating body will not be recognized by outside readers or some parsers. Expand less-common acronyms directly on first use.
Warning (dates): Date formatting is inconsistent in apostrophe and dash spacing (for example, "May'26 - present" versus "July' 25 - Apr'26"). Standardize to one format throughout.
Warning (formatting): Standalone footnote markers appear inside role content, interrupting bullets. These superscript numbers can create parsing artifacts. Replace endnotes with the spelled-out term directly in the relevant bullet.
Warning (keywords): Core PM keywords are present and well-placed in experience bullets (roadmap, stakeholder, strategy, metrics, prioritization, discovery, user research, requirements, go-to-market, impact, outcome). Underused: cross-functional, launch, iteration, trade-off, adoption. Several of these would appear naturally once the "what happened after" layer is added, since launch, adoption, and iteration are exactly that layer.
Key Takeaways
1. "Built" is where an engineer stops. A PM continues to "and then." The strongest bullets on this resume end at design or build. The PM version adds what happened next: who used it, how much it improved, what the stakeholder decided. The outcomes usually exist. Put them on the page.
2. The most important AI product bullet is the failure-mode one. Retrieval accuracy and latency show you can build. The guardrail for wrong answers (confidence thresholds, citations, escalation, human review) shows you can own an AI product. AI PM interviewers screen for the second.
3. Lead data projects with the decision, not the model metric. Macro-F1 and accuracy belong to the data science framing. The PM framing is: what user or business problem did this solve, who acts on the output, and what changes because of it. Reorder the bullet to put the decision first.
4. Breadth without a stated direction reads as unfocused. Five domains and no summary makes a recruiter assemble your story for you, and they often conclude "talented but scattered." Two lines of positioning turns the same breadth into intentional range.
5. A clean template unlocks the keyword gap for free. Footnote markers inside bullets and inconsistent dates hurt parsing. Fixing the layout, and adding the "what happened after" layer, brings in the missing launch, adoption, and iteration keywords naturally.
The Pattern
This is a "PM in all but title" resume from a strong undergrad whose product craft is already visible, held back by a single missing layer.
Unlike most student resumes, this one does not have a product-thinking problem. Discovery, requirements, prototypes, and roadmaps are all here in the candidate's own work. The gap is narrower and more fixable: the bullets stop at "designed" and "built" without showing what shipped, who used it, or what outcome moved. Add that layer to the top internships and the AI assistant, state a product direction up top, and this moves from an interested-but-conditional read to a clear interview yes.
Score your own resume to see how your product manager resume performs across all four dimensions.