# Content verification — Rico Shukla portfolio

Reviewed November 10, 2026. This record distinguishes user-supplied statements, repository documentation, inspected implementation, independently rerun tests, external research, and proposed features. Repository authors' study reports have not been independently audited against participant-level data.

## Person, experience, and contact

| Claim | Source | Treatment / limitations |
| --- | --- | --- |
| Rico Shukla; UT Dallas; B.S. Computer Information Systems Technology, Business Analytics and AI concentration; Finance minor; expected December 2027 | User's build brief | Published as user-supplied biographical information. |
| Current Product Management Intern, Watts Water Technologies | User's build brief | “Current experience”; no employment dates invented. |
| 50+ stakeholder interviews for a lower-cost commercial water heater controller | User's build brief | Self-reported experience; not independently verified with employer records. |
| Approximately 40 requirements developed and prioritized; MVP/roadmap presentation to leadership; Jira stories, acceptance criteria, backlog refinement | User's build brief | Scope is product work and collaboration; engineering implementation is not attributed to Rico. |
| Internal AI Product Intelligence Platform pilot with feedback and evaluation workflows | User's build brief | Called a pilot; no production adoption or performance results claimed. |
| 27–28% faster correct fault identification in technician study | User's build brief | Intentionally omitted from public highlights because study details are unavailable. |
| Product, engineering, analytics and collaboration toolkit | User's requested skills list | Presented as a toolkit, not certifications or quantified proficiency. Individual projects only claim source-confirmed technology. |
| GitHub: https://github.com/ricoxvl | User supplied; profile and repositories inspected | Used for code and project documentation. |
| LinkedIn: https://www.linkedin.com/in/rico-shukla/ | Explicitly supplied by Rico in the follow-up conversation | Used by “Let’s connect” in navigation and LinkedIn links in the hero and contact section. Profile content has not been independently audited. |

`site.config.json` contains nullable `linkedin`, `email`, and `resume` fields. LinkedIn is populated from the user-supplied URL. The résumé is the user-uploaded `Rico_Shukla_Resume (1).pdf`, served unchanged at `/Rico_Shukla_Resume.pdf` with download links in the homepage hero and contact sections. Email is not configured as a separate contact button. Résumé SHA-256: `9802de452de62df2107cc8b84952a3c1e3f630f72950fa6618864c2d989d1636`. Its inclusion does not supersede the explicitly corrected case-study figures. Add only verified values. No employer endorsement, company-logo asset, testimonial, or private company data is implied; WATTS is a typographic company identifier.

## Pathwise — private build, owner-confirmed; source verification pending

The user supplied https://github.com/ricoxvl/pathwise and said the public repository was empty while private work was being migrated. The initial public GitHub lookup returned 404. This does not establish that private implementation does not exist.

Rico subsequently supplied the full Pathwise product account in `Pasted text.txt` and explicitly confirmed that all described product capabilities are already built. The case study therefore uses **Private build**, replacing the earlier **Work in progress** label. Product scope and architecture are attributed to the owner's account. No private source, actual screen or implementation was independently inspected; the supplied narrative is not code-level verification.

Document SHA-256: `4147c9a0c341433c03a69b94f1c99dca5396c8707cc9db8e02b35a65167060b7`.

### What the supplied account supports

- Product positioning: academic risk intelligence and explainable decision support, with a narrow initial focus on selected UT Dallas undergraduate programs.
- Ten described product areas: academic onboarding, degree health, requirements explorer, prerequisite map, semester planner, what-if simulator, Ask Pathwise, advisor preparation, catalog comparison, and privacy/data control/sharing.
- Owner-reported product and technical contribution: product strategy, experience design and technical development. No additional team membership, employment connection or institutional endorsement is invented.
- Prioritization: requirements, prerequisites, risk dashboard and scenarios are core; source-grounded AI and advisor briefs are high priority; weekly scheduling is secondary; live seat alerts are deferred; automatic registration is excluded. These are the user's product decisions, not an empirical national ranking or fabricated numerical score.
- The documented architecture separates deterministic academic calculations, versioned retrieval and optional local explanations. Rule types include applicability, alternatives, minimum grades, course allocation, workload and administrative exceptions.
- The reported stack includes React/TypeScript/Vite, Tailwind/Lucide, graph utilities, versioned JSON, IndexedDB, PDF.js, Transformers.js and WebLLM. These are documented/owner-reported technologies, not source-inspected implementation claims.
- The $0 discussion is a **bounded operating-cost goal**, not verified hosting usage or an unlimited-cost guarantee. Device capability, downloads, browser retention, licenses and provider limits remain material trade-offs.
- A new “Graduation rescue journey” synthesizes the provided five-step scenario. Its Fall 2027 target, 12-credit limit, transfer uncertainty and course sequence are synthetic and do not constitute a real student's outcome or pilot result.

### Evidence limits preserved on the page

- The earlier source rule still applies: source-inspected implementation is distinguished from capabilities confirmed by Rico. The portfolio does not claim it inspected a functioning Pathwise deployment, codebase or ten actual screens.
- The pasted account contains both past-tense evaluation language and hypothetical targets, but **no measured results, reports, sample counts, test outputs or participant artifacts**. The follow-up confirms built capabilities, not numeric evaluation outcomes. Evaluation is therefore shown as a framework with no reported results; it is not asserted that evaluations definitely did or did not occur.
- Precision, recall, false all-clear errors, scenario validity, grounding, task completion, decision accuracy, calibration and advisor usefulness remain evaluation dimensions. Zero false all-clear errors is explicitly a target, not an achieved result.
- The north-star metric definition is the percentage of evaluated planning sessions with a validated actionable risk correctly identified before the academic decision. The website recommends reporting risk prevalence alongside it so risk-free sessions are not misread as failures. That recommendation is portfolio analysis, not a claimed implemented metric pipeline.
- Reddit discussion summaries, Orbit/PeterPortal/NC State comparisons and catalog-policy claims in the paste lack usable original source links. Domain labels and image-proxy URLs are not primary citations. They are not published as verified specific incidents, competitive facts, quotations or research counts. General qualitative themes are attributed to Rico's exploratory synthesis.
- The portfolio walkthrough remains a separate illustration with fictional FND 101, ANA 201 and CAP 301 courses. Its selector uses fixed examples; it is not Pathwise's academic engine or an actual UT Dallas dataset.
- The ten-area disclosure and the architecture include explicit source-verification context. There is no public runnable demo link until one can be confirmed.

### Original third-party research verified

1. [EDUCAUSE, ECAR Study of Undergraduate Students and Information Technology, 2019](https://library.educause.edu/-/media/files/library/2019/10/studentstudy2019.pdf?la=en&hash=25FBB396AE482FAC3B765862BA6B197DBC98B42C), p. 13: **80%** rated degree-audit tools very or extremely useful; **77%** rated degree-planning/mapping tools very or extremely useful. These figures concern surveyed users of their institution's success tools. They are not all-student percentages and are not original Pathwise research. Report quantitative findings use 40,596 responses across 118 US institutions, but the item-level bases for the tool-use figures must not be replaced with that overall denominator.
2. [Inside Higher Ed, “Academic Pathways: More Clarity Needed,” July 26, 2024](https://www.insidehighered.com/news/student-success/academic-life/2024/07/26/survey-students-want-more-clarity-around-academic): **28%** of 5,025 Student Voice respondents wanted their institutions to introduce or enhance online platforms for planning and tracking degree progress. The survey was conducted with Generation Lab. This is a requested institutional improvement, not Pathwise product-market fit or the prevalence of planning difficulty.

## CampusConnect — private build described by Rico; source review pending

Repository: https://github.com/ricoxvl/campusconnect

Inspected commit: `d9965c44e60ab188998648ff73a3025cbe72a1a4`.

[Primary README source](https://github.com/ricoxvl/campusconnect/blob/d9965c44e60ab188998648ff73a3025cbe72a1a4/README.md). The inspected public repository contains the README only. It reports a built application but publishes no application source, actual screenshots, raw research data, full protocol or runnable demo.

Rico also supplied the complete 21-section CampusConnect product account in `Pasted text.txt` (document SHA-256: `7452a53781123bbbc6564e82b11d9e02061c6d0834a3a0420b126686fb104458`) and requested its incorporation into the portfolio. The account describes the platform in completed-implementation terms. The site uses **Private build** with explicit owner-attribution and source-verification context; this does not establish a publicly launched production service.

### Corrected research figures

Rico explicitly confirmed **69/120 is accurate**. The case study uses the original README wording and corresponding percentage:

| Claim | Evidence and scope |
| --- | --- |
| 24 student interviews; 120 survey responses | README and expanded account; no participant-level data independently audited. |
| 67.5% had difficulty discovering relevant campus events | 81/120 respondents. Not a population prevalence estimate. |
| 63.3% reported recommendations missing their interests, schedules or goals | 76/120. The full original item wording is preserved; it is not reduced to interests alone. |
| 57.5% felt uncomfortable attending unfamiliar events | **69/120**, confirmed by Rico. The earlier README wording is preserved rather than equating unfamiliar-event discomfort with a separately worded alone-attendance question. |

### Expanded product account

The case study now explains the five systems **Discover, Plan, Connect, Organize and Understand** and three fictional journeys: transfer student, commuter and organization officer. It features Go Together, newcomer support, optional Campus Circles, commuter availability, club exploration, organizer publication, registration/check-in workflows, recommendation architecture, institutional reporting and commercial strategy.

- Rico's expanded role description includes product discovery, experience design and platform development. This is owner-reported; the public source does not establish individual engineering attribution or an additional team.
- React/TypeScript/Vite/Tailwind/Workers/D1 are documented in the README. Expanded AI, social, organizer, queue, security, monitoring, mobile and institutional capabilities are described in the supplied account, not independently code-reviewed.
- Go Together is described as event-specific mutual opt-in. The portfolio example does not contact anyone, create an invitation, reveal participants or access student schedules/locations. Joining and leaving change only the fictional example; switching personas resets it.
- Commuter dates/times and workshop examples are fictional. The portfolio has no live calendar, campus routing or event integration.
- Organizer artwork is a conceptual workflow, not a screenshot or operational publishing/check-in tool. No new attendance totals, users, customers or experimental outcomes are invented.
- The institutional SaaS section describes a commercial strategy. Signed university agreements, active paying customers, revenue and institutional endorsement are not established by the supplied artifacts.
- Responsive web, PWA and mobile clients are part of the owner's account; published iOS/Android releases are not independently verified and no app-store links are invented.
- Calendar and SSO integrations are described as optional/permissioned or authorization-dependent. Live deployments are not established. Accessibility practices are not presented as an independently audited WCAG-conformance result.

### Existing study results remain bounded

| Claim | Evidence and scope |
| --- | --- |
| 15 usability tests; task completion 70% → 90%; discovery time down 36% | README-reported comparisons across tested iterations. No randomized causal claim. Timing baselines, detailed protocol and feature/version mapping unavailable. |
| Earlier beta of 100 students | Total participation, separate from retention denominator. |
| 40% four-week retention | 30 active of 75 eligible, not 40/100. Full definitions and participant-level data unavailable. |

The expanded document does not identify which new capabilities were in the tested versions. These historical results are not generalized to Go Together, Campus Circles, native mobile apps, the AI ranking system or institutional workflows.

The north-star metric **Monthly Meaningful Participations per Active Student** is presented as a definition, without a reported score. Meaningful participation requires a supported, permitted confirmation; views, saves and RSVPs do not count as attendance. Measurement coverage and an independently specified active-student denominator should accompany it. Participation is not treated as proof of friendship or belonging. The denominator recommendation is portfolio analysis, not evidence of an implemented analytics pipeline.

## ApplyFlow — early coded MVP

Repository: https://github.com/ricoxvl/applyflow

Inspected commit: `114b32dd708ce0c9738b7842d9ae13b2b2212b0a`.

Sources:

- [README](https://github.com/ricoxvl/applyflow/blob/114b32dd708ce0c9738b7842d9ae13b2b2212b0a/README.md)
- [Application workflow](https://github.com/ricoxvl/applyflow/blob/114b32dd708ce0c9738b7842d9ae13b2b2212b0a/src/App.tsx)
- [Ranking and storage](https://github.com/ricoxvl/applyflow/blob/114b32dd708ce0c9738b7842d9ae13b2b2212b0a/src/logic.ts)
- [Fictional role catalog](https://github.com/ricoxvl/applyflow/blob/114b32dd708ce0c9738b7842d9ae13b2b2212b0a/src/catalog.ts)
- [Product direction](https://github.com/ricoxvl/applyflow/blob/114b32dd708ce0c9738b7842d9ae13b2b2212b0a/docs/product-direction-v02.md)
- [Original tracking-oriented brief](https://github.com/ricoxvl/applyflow/blob/114b32dd708ce0c9738b7842d9ae13b2b2212b0a/docs/product-brief.md)

Source inspection confirms React, TypeScript, Vite, editable experience/skills, role filters, explanations, required/preferred examples, saved roles, student-chosen actions/dates/notes, completion flags, localStorage and JSON export. Ranking is deterministic selected-skill overlap with alphabetical tie-breaking. Ten educational role archetypes are fictional. There is no validated AI matching algorithm, live job feed, placement prediction or required account/upload. A Google-search shortcut is not a live integration.

Discovery narratives are exploratory and owner-supplied. Participant details and consent are marked for reconciliation in the source. The tracking-to-discovery pivot is reported without invented quotes or statistical generalization. Small manually curated, self-reported test metrics in the direction document are intentionally omitted; they are not MVP performance results.

Older direction notes say “Figma concept,” but newer README and executable source establish a coded MVP. The current stage follows code evidence. This portfolio does not claim an independent ApplyFlow build or outcome study was completed.

Actual README Figma links:

- [v0.2 desktop](https://www.figma.com/design/AKLLIxOhHKBs8qdTUhKX2t?node-id=9-87)
- [v0.2 mobile](https://www.figma.com/design/AKLLIxOhHKBs8qdTUhKX2t?node-id=8-136)
- [Archived v0.1 tracking concept](https://www.figma.com/design/AKLLIxOhHKBs8qdTUhKX2t?node-id=1-18)

Exact URLs were verified in the repository. Independent HTTP access to Figma was blocked by the environment proxy, so current viewer access was not established. The portfolio labels them original artifacts and does not claim the prototypes were rendered or tested.

The portfolio skill-matching interaction is a separate teaching illustration with three fictional example roles. It does not call ApplyFlow, store a profile or show live jobs. Selections and saved-role actions reset on route change.

## FinPilot — local prototype with synthetic records

Repository: https://github.com/ricoxvl/ai-financial-operations-copilot

Inspected commit: `193b625a4fbf52f367ae85734432925fb75005e8`.

Sources:

- [README](https://github.com/ricoxvl/ai-financial-operations-copilot/blob/193b625a4fbf52f367ae85734432925fb75005e8/README.md)
- [PRD](https://github.com/ricoxvl/ai-financial-operations-copilot/blob/193b625a4fbf52f367ae85734432925fb75005e8/docs/PRD.md)
- [Architecture](https://github.com/ricoxvl/ai-financial-operations-copilot/blob/193b625a4fbf52f367ae85734432925fb75005e8/docs/architecture.md)
- [Evaluation notes](https://github.com/ricoxvl/ai-financial-operations-copilot/blob/193b625a4fbf52f367ae85734432925fb75005e8/docs/evaluation.md)
- [Committed offline evaluation report](https://github.com/ricoxvl/ai-financial-operations-copilot/blob/193b625a4fbf52f367ae85734432925fb75005e8/reports/offline_evaluation.json)

Python/FastAPI, SQLite/SQLAlchemy, validation, integer currency amounts, policy retrieval, duplicate checks, proposals, token-protected approval/rejection, idempotency and audit events were inspected in source. Optional LangGraph/OpenAI source exists, but live operation was not verified. The no-key path is a scripted deterministic demo, not a language model. Human approval creates a local invoice draft; nothing is sent, posted, or paid. Demo tokens are not production multi-user identity/RBAC.

### Independently rerun during this portfolio build

- `python -m pytest -q`: **52 passed**, 4.39 seconds; one Starlette TestClient deprecation warning.
- `python scripts/evaluate.py --output …`: **150/150 synthetic deterministic scenarios passed**.
- Rerun date: November 10, 2026. Base requirements installed in an isolated environment. No live API credentials or optional agent execution used.
- Offline categories: 20 customer lookup, 30 invoice search, 30 valid proposals, 25 duplicates, 20 policy, 15 insufficient-input, 10 oversized-amount cases.
- Tests cover approval/rejection, authorization checks, idempotency, duplicates, input validity, route smoke tests and simulated resilience/concurrency scenarios.
- These results are not live LLM accuracy, real production uptime, independent financial audits, banking reliability or customer-impact evidence.
- The React frontend's build and the live agent were not independently verified. No production financial integrations, payment processing, QuickBooks or real customer financial records are claimed.
- Historical metrics disclaimed by the repository are excluded.

The portfolio's approval interaction is a separate synthetic illustration. It has no backend, authorization service or connection to financial systems. It cannot issue an invoice or move money.

## Portfolio implementation and review limits

All project visuals are original code-rendered conceptual illustrations, explicitly labeled. No fabricated screenshot, testimonial or participant quotation is used. No source project had image assets to reuse. Page text, sources and dates are maintained in `src/data/projects.ts`; contact details in `site.config.json`.

The website provides static prerendered React routes with client hydration, unique metadata, Open Graph imagery, a sitemap, responsive CSS, reduced-motion handling and functional navigation. Tests cover routes, interaction decisions, metadata, exported links and build assets. These are portfolio checks, separate from the upstream product evaluation evidence above.

Browser layout/visual QA could not be completed in this environment: the managed desktop reports enabled, but its `cua_repl.js` browser-control tool is unavailable. The Sites workflow prohibits substituting another browser-control path in that condition. No screenshot, real-browser accessibility audit or device-tested claim is made. Run the documented visual release checklist before a broad recruiter launch.