[{"content":"What It Does A conversational AI chatbot specifically configured for HR questions. Employees can ask questions about CAO (collective labor agreement), leave entitlements, sick leave procedures and get instant, accurate answers at any time of day. The tool provides 24/7 support without requiring HR staff intervention for routine policy questions.\n🚀 Launch Live Demo\nThe Problem It Solves HR teams spend a large portion of their time answering repetitive policy questions. This constant context-switching diverts strategic HR work and leaves employees waiting for answers. When HR is unavailable, employees struggle to find clear answers to common questions about their rights, entitlements, and procedures. This tool offloads those routine queries so HR can focus on strategic work while employees get instant answers whenever they need them.\nHow It Works The tool uses the Claude API (routed via a Cloudflare Worker proxy) to power a branded chat interface. Responses are grounded in Kust \u0026amp; Kade HR policy context passed in the system prompt. The system ensures consistency and accuracy by providing the AI with a complete HR policy reference, so answers are always aligned with company standards.\nKey Features 24/7 instant answers: No waiting for HR availability Conversational chat interface: Natural, multi-turn conversation with message history Branded Kust \u0026amp; Kade design: Fully customized with company colors and tone of voice Mobile-friendly layout: Works seamlessly on any device AI-powered with Claude: Leverages Claude\u0026rsquo;s natural language understanding for accurate HR guidance Tech Stack Layer Technology Frontend HTML/CSS/JavaScript (vanilla) API Claude API Infrastructure Cloudflare Workers (proxy) Typography Google Fonts ","permalink":"https://0xthijs.github.io/projects/kust-kade-hr-assistent/","summary":"An AI-powered HR assistant that gives instant, 24/7 answers to employee questions about collective labor agreements, leave, and absence policies — built for Kust \u0026amp; Kade.","title":"🤖 HR Assistent — Kust \u0026 Kade"},{"content":"What It Does Employees or L\u0026amp;D coaches enter a development goal and the tool generates a structured 70-20-10 action plan: 70% on-the-job learning, 20% social/peer learning, 10% formal training — all concretely tailored to the goal. The result is an immediately actionable personal development roadmap.\n🚀 Launch Live Demo\nThe Problem It Solves Personal development plans (POPs) are often vague and quickly forgotten. Managers and employees struggle to translate intentions into concrete, measurable actions. The 70-20-10 learning model is powerful but requires expertise to apply. This tool bridges that gap by automatically structuring development goals into actionable, balanced learning plans that align with proven adult learning science.\nHow It Works The user\u0026rsquo;s development goal is sent to the Claude API (via a Cloudflare Worker proxy) with a prompt designed around the 70-20-10 learning model. The AI returns a structured, actionable plan per category. Each section includes specific activities, resources, and milestones aligned to the employee\u0026rsquo;s learning style and organizational context.\nKey Features Input any development goal: Natural language description of what the employee wants to learn or improve AI-generated 70-20-10 structured action plan: Automatically balances learning approaches Clear breakdown across on-the-job / social / formal learning: Easy to understand and execute Branded Kust \u0026amp; Kade design: Professional, on-brand presentation Mobile-friendly: Full functionality on phones, tablets, and desktops Tech Stack Layer Technology Frontend HTML/CSS/JavaScript (vanilla) API Claude API Infrastructure Cloudflare Workers (proxy) Typography Google Fonts ","permalink":"https://0xthijs.github.io/projects/kust-kade-ontwikkelkompas/","summary":"An AI tool that transforms vague personal development goals into concrete 70-20-10 action plans — built for Kust \u0026amp; Kade\u0026rsquo;s learning \u0026amp; development practice.","title":"🧭 Ontwikkelkompas — Kust \u0026 Kade"},{"content":"What It Does HR or a manager fills in details about a new hire (role, team, start date) and instantly gets a structured 4-week onboarding plan with weekly goals, tasks, and check-in moments. The plan is personalized to the specific role and team, making every new employee\u0026rsquo;s first month clear and structured.\n🚀 Launch Live Demo\nThe Problem It Solves Onboarding is often improvised and inconsistent. New hires frequently feel lost in the first weeks, unsure what to focus on or when they should have mastered key skills. Managers struggle to balance getting employees productive while ensuring proper integration. This tool gives every new employee a clear, personalized roadmap from day one, reducing anxiety, improving time-to-productivity, and ensuring consistency across the organization.\nHow It Works Pure client-side tool — no API calls required. The logic runs entirely in the browser, generating a structured week-by-week plan based on the inputs provided (role, team, department, start date). The tool is fully offline-capable and requires no backend infrastructure, making it instantly available and completely private.\nKey Features 4-week structured onboarding plan: Organized week-by-week with clear progression Personalized per role and team: Adapts to specific job context and department needs No API or backend required: Runs fully in-browser, completely offline-capable Printable / shareable output: Easy to distribute to new hire and team members Branded Kust \u0026amp; Kade design: Consistent with company identity and professional appearance Tech Stack Layer Technology Frontend HTML/CSS/JavaScript (vanilla) Architecture Client-side only (no backend) Infrastructure None required (fully offline) Typography Google Fonts ","permalink":"https://0xthijs.github.io/projects/kust-kade-onboarding-routekaart/","summary":"A static interactive tool that generates a personalised 4-week onboarding roadmap for new employees — practical, visual, and ready to use on day one.","title":"🗺️ Onboarding Routekaart — Kust \u0026 Kade"},{"content":"What It Does HR Analytics 2030 is a secure, local-first dashboard that transforms raw HRIS data into actionable workforce intelligence. Unlike traditional BI tools that require cloud uploads, this application runs entirely in your browser using IndexedDB and Client-Side Logic, ensuring sensitive employee data never leaves your device.\n🚀 Launch Live Dashboard\nThe Problem It Solves HR teams often struggle with:\nData Privacy: Hesitancy to upload sensitive payroll/attrition data to public cloud tools or external SaaS vendors. Static Reporting: Relying on backward-looking Excel sheets instead of forward-looking scenarios. Complex Modeling: Lack of accessible tools to model \u0026ldquo;what-if\u0026rdquo; scenarios for attrition and retirement without a data science team. Architecture \u0026amp; How It Works The application uses a Local-First Architecture. The browser is the database.\ngraph TD User[HR User] --\u0026gt;|Uploads CSV| UI[Next.js UI] UI --\u0026gt;|Parses \u0026amp; Sanitizes| Worker[Client Logic] Worker --\u0026gt;|Stores Data| DB[(IndexedDB / Dexie.js)] subgraph \u0026#34;Browser Sandbox (Secure)\u0026#34; DB --\u0026gt;|Query Employee Data| Engine[Heuristic Engine] Engine --\u0026gt;|Detect| Risk[Attrition \u0026amp; Retirement Risks] Engine --\u0026gt;|Calculate| Plan[Workforce Plan] Risk --\u0026gt;|Visualize| Charts[Recharts Dashboard] Plan --\u0026gt;|Visualize| Charts end Charts --\u0026gt; User Key Features 🔒 Privacy-First Design Zero data exfiltration. The application uses IndexedDB to persist data strictly within your browser profile. You can upload a 10,000-row CSV, close the tab, come back later, and your data is still there—but it never touched a server.\n🔮 Predictive Attrition Modeling The system runs a heuristic analysis on every employee to flag High Flight Risks based on:\nTenure Cliff: Detecting employees in the critical 2-4 year turnover window. Compa-Ratio: Identifying high performers (Rating 4-5) with below-market salaries. Retirement Risk: Flagging key roles approaching retirement age. 📈 Interactive Scenario Planning Stop guessing. The \u0026ldquo;Plan\u0026rdquo; module allows leaders to toggle a single \u0026ldquo;Growth Target\u0026rdquo; slider (e.g., +15% YoY) and immediately see:\nHow many new hires are needed per department. The gap between current headcount, attrition, and future targets. Tech Stack Layer Technology Framework Next.js 15 (App Router, Static Export) Language TypeScript Database Dexie.js (IndexedDB Wrapper) Styling Tailwind CSS v4 Visualization Recharts Parsing PapaParse (Stream-capable CSV parsing) View Source Code\n","permalink":"https://0xthijs.github.io/projects/hr-analytics-2030/","summary":"A privacy-first predictive analytics platform. Features client-side AI for attrition modeling, interactive workforce planning scenarios, and secure CSV processing—all running 100% in your browser.","title":"📊 HR Analytics 2030"},{"content":"What It Does SkillFlex is an AI-driven internal talent marketplace that automatically matches employees to short-term projects (\u0026ldquo;gigs\u0026rdquo;) based on their inferred skills. It replaces static, manual skill profiles with dynamic, AI-generated talent signals to uncover hidden potential within the workforce.\nThe Problem It Solves High-performing employees often leave diverse organizations because they cannot find internal growth opportunities, while managers struggle to find talent for short-term needs. Traditional HR systems rely on manual data entry, resulting in outdated skills inventories that fail to capture an employee\u0026rsquo;s true capabilities or flight risk.\nHow It Works The system uses a Google Gemini Pro agent to ingest unstructured data (resumes, project history, Slack/Teams activity) and infer a \u0026ldquo;Live Skills\u0026rdquo; profile for each employee. It then uses a retention-weighted algorithm to match these profiles against open internal gigs, prioritizing matches that reduce attrition risk for high-value talent.\ngraph TD A[Employee Profile] --\u0026gt;|Unstructured Data| B(Ingestion Pipeline) B --\u0026gt;|Context Window| C{Gemini Pro Agent} C --\u0026gt;|Structured JSON| D[Skills Database] E[Manager Gigs] --\u0026gt;|Requirements| F[Matching Engine] D --\u0026gt; F F --\u0026gt;|Weighted Jaccard Algo| G[Opportunity Feed] subgraph \u0026#34;AI Logic Layer\u0026#34; C F end subgraph \u0026#34;Retention Safeguard\u0026#34; H[Attrition Risk Flags] -.-\u0026gt;|Boost Score| F end Key Features AI Skill Inference: Automatically extracts 5 hard and 3 soft skills from unstructured text using LLMs, removing the need for manual profile updates. Retention-Weighted Matching: The matching engine applies a 1.2x score multiplier to employees flagged as \u0026ldquo;High Flight Risk,\u0026rdquo; surfacing opportunities to retain them. Premium SaaS Experience: A fully responsive, dark-mode enabled UI (built with Flask \u0026amp; Tailwind) that provides a consumer-grade experience for enterprise users. Explainable AI: Every gig recommendation includes a \u0026ldquo;Why this matches you\u0026rdquo; section to build user trust in the AI suggestions. Results / Impact This Proof of Concept demonstrates how Generative AI can transform internal mobility from a passive database into an active retention engine. By inferring skills rather than asking for them, SkillFlex reduces the friction of profile creation to zero, promoting higher adoption rates than traditional manual systems.\nTech Stack Layer Technology Frontend Flask (Jinja2) / TailwindCSS / Alpine.js Backend Python / Flask AI Engine Google Gemini Pro Data SQLAlchemy / SQLite (PoC) Deployment Localhost (Demo) View Code\n","permalink":"https://0xthijs.github.io/projects/skillflex/","summary":"An AI-native marketplace that infers employee skills from unstructured data to automatically match talent with internal projects, reducing attrition.","title":"🧠 SkillFlex"},{"content":"What It Does The Automated Contract Generator is a client-side tool that transforms standardized HR templates into dynamic, interactive forms. It allows HR generalists to draft complex employment agreements by simply filling out a structured questionnaire, with changes reflected instantly in a print-ready document.\nThe Problem It Solves Drafting employment contracts manually using \u0026ldquo;Find \u0026amp; Replace\u0026rdquo; is error-prone and time-consuming. Legal teams often struggle to enforce version control, leading to outdated clauses being sent to candidates.\nHow It Works The application uses Vanilla JavaScript to create a reactive DOM binding system. It maps input fields directly to span elements within the contract text, updating content in real-time as the user types. The system runs entirely in the browser, ensuring candidate data privacy by never sending PII to a server.\nKey Features Dynamic Template Injection: Text fields and variables update instantly across the entire document as inputs change. Smart Formatting: Automatically handles currency localization (e.g., converting inputs to $120,000) and date formats. Privacy-First Architecture: Zero backend dependency means sensitive candidate data never leaves the user\u0026rsquo;s device. Results / Impact Reduces contract drafting time from 20 minutes to under 2 minutes per candidate while eliminating formatting errors and ensuring 100% compliance with the current legal template.\nTech Stack Layer Technology Frontend HTML5 / CSS3 / Vanilla JS Logic DOM Manipulation API Deployment GitHub Pages View Code\n","permalink":"https://0xthijs.github.io/projects/automated-contract-generator/","summary":"A browser-based engine that generates error-free employment contracts in real-time, eliminating manual drafting mistakes.","title":"📜 Automated Contract Generator"},{"content":"What It Does The Burnout Risk Model is a composite scoring system that quantifies \u0026ldquo;hidden\u0026rdquo; employee stress factors. It aggregates distinct data points—such as commute distance, overtime hours, and promotion stagnation—into a single 0-100 \u0026ldquo;Burnout Score\u0026rdquo; for every employee.\nThe Problem It Solves Burnout is often treated as a qualitative feeling, leaving HR leaders reactive rather than proactive. By the time an employee reports feeling burnt out in a survey, they are often already looking for a new job.\nHow It Works I engineered a Composite Risk Score using Python (Pandas) to weigh and aggregate stressors found in the IBM HR dataset. The model assigns weighted points to risk factors:\nOvertime (+30 pts) Commute (\u0026gt;20 miles) (+20 pts) Stagnation (No promotion \u0026gt;4 years) (+10 pts) Sentiment (Low Work-Life Balance score) (+40 pts) Key Features Weighted Scoring Algorithm: Differentiates between minor irritants and major flight risks. Risk Segmentation: Automatically categorizes the workforce into \u0026ldquo;Safe,\u0026rdquo; \u0026ldquo;At Risk,\u0026rdquo; and \u0026ldquo;Critical\u0026rdquo; cohorts. Visual Impact: Generates clear distribution charts to highlight organizational hotspots. Results / Impact The model successfully identified a \u0026ldquo;Red Zone\u0026rdquo; cohort of 64 employees who were 4x more likely to resign (39.1% attrition rate) compared to the low-risk group (9.4%). This allowed HR to deploy targeted retention interventions before resignations occurred.\nTech Stack Layer Technology Analysis Python (Pandas / NumPy) Visualization Matplotlib / Seaborn Data Source IBM HR Analytics Dataset ","permalink":"https://0xthijs.github.io/projects/burnout-risk/","summary":"A predictive risk model that aggregates commute, overtime, and stagnation data to identify potential attrition, proving 4x more effective than random sampling.","title":"🔥 Burnout Risk Model"},{"content":"What It Does CompSense is a specialized dashboard for Compensation \u0026amp; Benefits professionals that transforms static salary data into a dynamic planning environment. It allows HR teams to model pay raise scenarios, visualize market positioning, and distribute merit increases while staying within budget.\nThe Problem It Solves Traditional compensation planning relies on complex, error-prone spreadsheets that lack visual context. Managers struggle to see the immediate budgetary impact of their decisions or how an employee\u0026rsquo;s new salary compares to market bands.\nHow It Works The application is built with React 19 and TypeScript for type-safe, component-based architecture. It uses Recharts to render interactive scatter plots and data grids that update instantly as variables (like global merit increase %) are adjusted.\nKey Features Pay Raise Simulator: Interactive slider that models different merit increase percentages and their impact on the total payroll budget. Market Position Analysis: Visual scatter plot comparing employee salaries against market bands (Junior to Principal), color-coded by role. Real-Time Data Grid: Detailed employee table that recalculates projected salaries and Compa-Ratios instantly. Smart Tooltips: Context-aware popovers showing projected salary and market alignment on hover. Results / Impact Enables accurate, data-driven compensation reviews by providing immediate visual feedback on budget utilization and market equity.\nTech Stack Layer Technology Framework React 19 / TypeScript / Vite Styling Tailwind CSS v4 Visualization Recharts Deployment GitHub Pages View Code\n","permalink":"https://0xthijs.github.io/projects/compsense/","summary":"A modern, interactive compensation planning dashboard featuring real-time budget modeling and salary band visualization.","title":"💰 CompSense"},{"content":"What It Does Crystal Onboarding is a highly interactive, premium onboarding experience designed to make the crucial first steps of a new hire feel less like filling out forms and more like an engaging, personalized journey.\n🚀 Launch Live Demo\nThe Problem It Solves Standard enterprise onboarding is often a cold, transactional process of clicking checkboxes. This project reimagines the experience into a visually stunning, dynamic flow that reduces anxiety and builds anticipation for the first day.\nHow It Works The application uses React Three Fiber to render a 3D crystal that visually evolves as the user progresses through the onboarding phases. Depending on their choices (hardware, work style, birth year, skills), the crystal changes its color, energy lines, and subtle zodiac glow, ultimately revealing their unique professional \u0026ldquo;avatar.\u0026rdquo;\nKey Features Dynamic 3D Visualization: A beautiful wireframe crystal that responds to user input. Vogue Aesthetic: A premium, dark-themed, minimalist UI that centers on elegant interaction. Gamified Progression: Breaking down hardware setup, account creation, and team introductions into bitesize, rewarding steps. Instant Colleague Connection: Built-in interactive cards to send the first \u0026ldquo;hello\u0026rdquo; messages to key team members. Tech Stack Layer Technology Frontend Framework Next.js / React Styling Tailwind CSS 3D Rendering React Three Fiber / Three.js State Management Zustand Animations Framer Motion Deployment Webpack Static Export (GitHub Pages) View Code\n","permalink":"https://0xthijs.github.io/projects/crystal-onboarding/","summary":"A premium, gamified onboarding journey featuring a personalized, interactive 3D crystal visualization that adapts to the new hire\u0026rsquo;s profile.","title":"🔮 Crystal Onboarding"},{"content":"What It Does The Critical Talent Risk Radar is an HR analytics application that segments the workforce to identify \u0026ldquo;Critical Talent\u0026rdquo; (High Potential / High Impact) who are at risk of leaving. It then uses Generative AI to act as a retention consultant, drafting specific strategies for each risk case.\nThe Problem It Solves Organizations often fail to identify flight risks among their top performers until it\u0026rsquo;s too late. Generic retention programs fail to address the specific needs of critical talent, leading to the loss of key institutional knowledge.\nHow It Works The app uses Streamlit for the frontend and Pandas for data processing. It filters employee data based on performance, tenure, and sentiment scores to flag high-risk individuals. The Google Gemini API is then engaged to analyze the profile of at-risk departments and generate tailored retention advice.\nKey Features Automated Risk Segmentation: Instantly filters employees based on multi-dimensional criteria (Performance vs. Flight Risk). Executive Dashboard: Real-time visualization of critical talent exposure across departments. AI Retention Consultant: Generates contextual, actionable retention plans for specific risk cohorts. Results / Impact Transforms retention from a reactive guessing game into a proactive, targeted strategy, specifically protecting the organization\u0026rsquo;s most valuable human assets.\nTech Stack Layer Technology Frontend Streamlit Data Processing Python (Pandas) Visualization Plotly Express AI Engine Google Gemini API View Code\n","permalink":"https://0xthijs.github.io/projects/critical-talent-radar/","summary":"An AI-powered retention tool that identifies at-risk high performers and generates personalized retention strategies.","title":"🛡️ Critical Talent Risk Radar"},{"content":"What It Does This automated \u0026ldquo;Governance Engine\u0026rdquo; acts as a firewall for HR data, ensuring 100% integrity before data enters the analytics warehouse. It uses a \u0026ldquo;Chaos Engineering\u0026rdquo; approach, intentionally injecting errors to prove the system\u0026rsquo;s ability to catch them.\nThe Problem It Solves HRIS migrations and analytics projects often fail due to \u0026ldquo;dirty data\u0026rdquo;—human errors like typos (e.g., negative ages), impossible tenure dates, or orphaned records. Manual auditing of thousands of rows is slow, expensive, and error-prone.\nHow It Works The pipeline is built in Python using Pandas. It implements a two-stage process:\nChaos Monkey: Intentionally corrupts a sample of clean data with common HR errors (typos, logic conflicts). Audit Engine: A strict validation layer that must catch 100% of the injected errors to pass the build. Key Features Chaos Monkey Simulation: Stresses the system by injecting random logic errors (e.g., \u0026ldquo;Start Date \u0026gt; End Date\u0026rdquo;). Strict Schema Validation: Enforces business logic rules (e.g., \u0026ldquo;Director level must have \u0026gt;5 years tenure\u0026rdquo;). Executive Health Scorecard: Automatically generates a \u0026ldquo;Data Health\u0026rdquo; report with actionable cleanup tasks. Results / Impact Achieved 100% data trust for the People Analytics team by catching critical data quality issues before they polluted the dashboard, saving an estimated 40 hours of manual cleanup per month.\nTech Stack Layer Technology Logic Python / Pandas Testing PyTest (Chaos Monkey) Reporting Markdown / Pandas Profiling Deployment Local Script / CI Pipeline View Code\n","permalink":"https://0xthijs.github.io/projects/data-governance/","summary":"An automated \u0026lsquo;Governance Engine\u0026rsquo; acting as a firewall for HR data, implementing a \u0026lsquo;Chaos Monkey\u0026rsquo; to stress-test data integrity.","title":"🏗️ 100% Data Integrity Pipeline"},{"content":"What It Does This project is a comprehensive audit of workforce diversity data, designed to answer two critical questions: \u0026ldquo;Do we pay fairly?\u0026rdquo; (Pay Equity) and \u0026ldquo;Do we promote equitably?\u0026rdquo; (Opportunity Equity).\nThe Problem It Solves DEI initiatives often rely on good intentions rather than hard data. Without granular measurement, organizations cannot identify if their diversity gaps are caused by bias in hiring, paying, or promoting.\nHow It Works I used Python to calculate the Adjusted Pay Gap, comparing average monthly income between genders within the same job level to isolate the effect of gender from seniority. I also mapped the \u0026ldquo;Representation Funnel\u0026rdquo; to visualize the drop-off rates at each management tier.\nKey Features Adjusted Pay Gap Analysis: Isolates pay discrepancies by controlling for role and level. Pipeline Visualization: Sankey-style analysis of the \u0026ldquo;Leaky Pipeline\u0026rdquo; to the C-Suite. Representation Heatmap: Identifies specific departments lacking diversity. Results / Impact Pay Equity Confirmed: Validated that the adjusted pay gap is \u0026lt; 2% across all levels. Leaky Pipeline Identified: Revealed a significant drop in female representation from Director (48%) to Executive (34%), shifting the strategy to focus on internal sponsorship rather than just hiring. Tech Stack Layer Technology Analysis Python (Pandas) Visualization Matplotlib / Seaborn Data HRIS Snapshot ","permalink":"https://0xthijs.github.io/projects/diversity-audit/","summary":"A rigorous audit of the Gender Pay Gap and Leadership Representation, uncovering a \u0026rsquo;leaky pipeline\u0026rsquo; despite achieving pay equity.","title":"🌈 Diversity Audit 2026"},{"content":"What It Does This analysis investigates the hidden drivers of employee turnover using the IBM HR Attrition dataset. It tests common retention hypotheses against statistical reality to determine what actually makes people stay.\nThe Problem It Solves Organizations often default to \u0026ldquo;throwing money at the problem\u0026rdquo; (salary hikes) or assuming high performers are safe. This analysis proves which levers actually move the needle on retention.\nHow It Works I performed exploratory data analysis (EDA) using Python (Pandas) to correlate attrition rates with variables like salary, manager tenure, and performance ratings.\nKey Findings High Performer Risk: Employees with a 4/5 rating had a 16.37% attrition rate (higher than the 16.12% average). The \u0026ldquo;New Manager\u0026rdquo; Cliff: Attrition spikes to 32% during the first year with a new manager. Salary Hikes Ineffective: The average salary hike for leavers (15.10%) was virtually identical to stayers (15.23%), debunking the \u0026ldquo;pay to stay\u0026rdquo; myth. Results / Impact Shifted the retention strategy from broad salary bands to targeted \u0026ldquo;New Manager Integration\u0026rdquo; programs and \u0026ldquo;Stay Interviews\u0026rdquo; for high performers, directly addressing the highest-risk segments.\nTech Stack Layer Technology Analysis Python (Pandas) Visualization Matplotlib Data Source IBM HR Analytics Dataset ","permalink":"https://0xthijs.github.io/projects/ibm-attrition/","summary":"A deep-dive analysis challenging retention myths, revealing that \u0026lsquo;High Performers\u0026rsquo; are 16% more likely to leave than average employees.","title":"🏃 High Performer Attrition Analysis"},{"content":"What It Does The ONA Visualizer maps the \u0026ldquo;Invisible Organization\u0026rdquo;—the informal web of communication that actually drives work. It visualizes connections between employees to reveal who is truly influential and who is becoming isolated.\nThe Problem It Solves Traditional org charts only show reporting lines, masking the reality of how collaboration happens. This leads to blind spots where key influencers burn out unnoticed, or remote teams become siloed from the core business.\nHow It Works The tool uses D3.js to render a force-directed graph where nodes represent employees and links represent communication frequency. It serves as a privacy-first, client-side PoC that runs directly in the browser without backend data storage.\nKey Features Network Strength Scoring: Identifies \u0026ldquo;Hubs\u0026rdquo; (informal leaders) whose departure would fracture the network. Proximity Bias Detection: Visualizes the disconnect between \u0026ldquo;Remote\u0026rdquo; (Purple) and \u0026ldquo;In-Office\u0026rdquo; (Green) clusters. Retention Risk Analysis: Highlights critical nodes that are \u0026ldquo;single points of failure\u0026rdquo; for information flow. Results / Impact Visualized the isolation of the \u0026ldquo;Remote\u0026rdquo; cluster, providing data-backed evidence to support a new hybrid integration program.\nTech Stack Layer Technology Frontend HTML5 / CSS3 Visualization D3.js v7 Logic Vanilla JavaScript Deployment GitHub Pages View Code\n","permalink":"https://0xthijs.github.io/projects/ona-visualizer/","summary":"A browser-based Organizational Network Analysis tool that visualizes informal communication flows to spot hidden influencers and isolated teams.","title":"🕸️ ONA Visualizer"},{"content":"What It Does This is a lightweight, purposeful Proof-of-Concept for an interactive employee onboarding checklist. It allows new hires to track their first-week tasks through a gamified interface that saves progress locally on their device, requiring no login or installation.\nThe Problem It Solves Enterprise onboarding tools are often heavy, slow, and require complex provisioning before a user can even log in. This creates a \u0026ldquo;chicken and egg\u0026rdquo; problem where the new hire needs the tool to get access to the tool.\nHow It Works The application leverages the browser\u0026rsquo;s LocalStorage API to persist user data without a database. It is built with Vanilla JavaScript to ensure instant load times and zero dependency vulnerabilities.\nKey Features Local Persistence: Progress is saved to the browser, surviving page reloads or browser restarts. Zero-Friction Access: No login, no servers, no installation—just a URL. Gamified Progress: Visual completion bars and \u0026ldquo;celebration\u0026rdquo; states to drive user engagement. Results / Impact Demonstrated the viability of \u0026ldquo;Micro-Apps\u0026rdquo; for HR: lightweight tools that solve specific problems without adding to the enterprise tech bloat.\nTech Stack Layer Technology Frontend HTML5 / CSS3 Logic Vanilla JavaScript Persistence LocalStorage API Deployment GitHub Pages View Code\n","permalink":"https://0xthijs.github.io/projects/onboarding-poc/","summary":"A zero-dependency, local-first onboarding checklist that gamifies the new hire experience without requiring any backend infrastructure.","title":"👋 Zero-Dependency Onboarding"},{"content":"What It Does \u0026ldquo;The 1:1 Architect\u0026rdquo; is a productivity tool that helps employees prepare for bi-weekly manager meetings. It prompts users to input their raw thoughts and unstructured updates, then uses simple logic (and planned AI integration) to reformat them into a high-impact agenda.\nThe Problem It Solves Manager meetings often devolve into \u0026ldquo;status updates\u0026rdquo; or casual chats, leaving no time for meaningful career discussions. Employees struggle to articulate their wins or frame their blockers strategically.\nHow It Works Built on Streamlit, the app provides a guided workflow that forces the user to categorize their updates into \u0026ldquo;Strategic Wins,\u0026rdquo; \u0026ldquo;Critical Blockers,\u0026rdquo; and \u0026ldquo;Discussion Topics.\u0026rdquo; It allows users to toggle between \u0026ldquo;Direct\u0026rdquo; and \u0026ldquo;Diplomatic\u0026rdquo; output modes to suit their manager\u0026rsquo;s style.\nKey Features Structured Agenda Generation: Automatically formats inputs into a clear, readable briefing document. Tone Calibration: Offers \u0026ldquo;Direct\u0026rdquo; vs. \u0026ldquo;Diplomatic\u0026rdquo; phrasing options for sensitive topics. Privacy-First: Runs locally to ensure sensitive career discussions remain private. Results / Impact Empowers employees to take ownership of their management relationship, shifting the focus from \u0026ldquo;what I did\u0026rdquo; to \u0026ldquo;what I achieved.\u0026rdquo;\nTech Stack Layer Technology Frontend Streamlit Logic Python deployment Localhost View Code\n","permalink":"https://0xthijs.github.io/projects/one-on-one-architect/","summary":"A Streamlit tool designed to restructure manager meetings, transforming unstructured chats into strategic career conversations.","title":"🗣️ The 1:1 Architect"},{"content":"What It Does This project tests the \u0026ldquo;Stagnation Hypothesis\u0026rdquo;—the idea that people leave because they aren\u0026rsquo;t moving up. Using the IBM HR dataset, I analyzed the attrition rates of high potential employees based on their time since last promotion.\nThe Problem It Solves Companies often view promotions as a \u0026ldquo;Retention Lock,\u0026rdquo; assuming a promoted employee is safe for at least 2 years. This false sense of security leads to a lack of support during the critical transition period.\nHow It Works I performed a cohort analysis using Python (Pandas), segmenting high performers into \u0026ldquo;Stagnant\u0026rdquo; (\u0026gt;2 years without promotion) and \u0026ldquo;Propelled\u0026rdquo; (\u0026lt;2 years since promotion) groups and calculating their respective flight risk.\nKey Findings The Promotion Paradox: \u0026ldquo;Propelled\u0026rdquo; high performers had a 17.0% attrition rate, compared to 13.7% for their \u0026ldquo;Stagnant\u0026rdquo; peers. Title Shopping: New titles make employees significantly more marketable to external recruiters. The Valley of Despair: The stress of a new role combined with \u0026ldquo;Mission Accomplished\u0026rdquo; syndrome increases vulnerability to poaching. Results / Impact Reframed the internal mobility strategy to include \u0026ldquo;Post-Promotion Onboarding\u0026rdquo; programs, treating internal movers with the same care and support structure as new external hires.\nTech Stack Layer Technology Analysis Python (Pandas) Visualization Matplotlib Data IBM HR Analytics Dataset ","permalink":"https://0xthijs.github.io/projects/promotion-velocity/","summary":"A counter-intuitive analysis of the \u0026lsquo;Promotion Paradox\u0026rsquo;, revealing that high performers are 17% more likely to leave immediately after a promotion.","title":"🚀 The Promotion Curse"},{"content":"Executive Summary This quarter, the HR Data Lab focused on Retention Mechanics. We analyzed the lifecycle of our high performers to answer one question: Why do people leave?\nThe Answer: It is not pay. It is Transitions and Burnout.\ngraph TD A[Q1 2026 HR Strategy] --\u0026gt; B{Key Risks} B --\u0026gt; C(Talent Risk) B --\u0026gt; D(Operational Risk) B --\u0026gt; E(Compliance) C --\u0026gt; C1[Promotion Curse] C1 --\u0026gt; C2[\u0026#34;17% Attrition \u0026lt;br/\u0026gt;(Recently Promoted)\u0026#34;] style C2 fill:#ffcccc,stroke:#333,stroke-width:2px D --\u0026gt; D1[Burnout Risk] D1 --\u0026gt; D2[\u0026#34;39% Attrition \u0026lt;br/\u0026gt;(High Risk Segment)\u0026#34;] style D2 fill:#ff9999,stroke:#333,stroke-width:2px E --\u0026gt; E1[Diversity Health] E1 --\u0026gt; E2[\u0026#34;Pay Gap \u0026lt; 2% \u0026lt;br/\u0026gt;(Target Achieved)\u0026#34;] style E2 fill:#ccffcc,stroke:#333,stroke-width:2px 📌 Critical Risks identified 1. The \u0026ldquo;Promotion Curse\u0026rdquo; (Talent Risk) We traditionally view promotions as a retention tool. Our data proves the opposite.\nFinding: High Performers promoted in the last 2 years are 17.0% likely to leave, vs 13.7% for those who are stagnant. Rec: Launch \u0026ldquo;Post-Promotion Onboarding\u0026rdquo; program immediately. 2. The \u0026ldquo;Red Zone\u0026rdquo; (Operational Risk) We successfully built a predictive model for burnout using existing data fields (Overtime + Commute + Role).\nFinding: 64 Employees are in the \u0026ldquo;High Risk\u0026rdquo; category with a 39.1% attrition rate (4x the norm). Rec: Immediate workload audit for these 64 individuals. 3. Diversity Health Check (Compliance) Status: ✅ Green. Adjusted Pay Gap is negligible (\u0026lt;2%). Watchlist: Executive representation (Level 5) remains low at 34%. For deep dives, please refer to the individual Project Case Studies linked in the Portfolio.\n","permalink":"https://0xthijs.github.io/reports/q1-2026-board-update/","summary":"\u003ch2 id=\"executive-summary\"\u003eExecutive Summary\u003c/h2\u003e\n\u003cp\u003eThis quarter, the HR Data Lab focused on \u003cstrong\u003eRetention Mechanics\u003c/strong\u003e. We analyzed the lifecycle of our high performers to answer one question: \u003cem\u003eWhy do people leave?\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eThe Answer:\u003c/strong\u003e It is not pay. It is \u003cstrong\u003eTransitions\u003c/strong\u003e and \u003cstrong\u003eBurnout\u003c/strong\u003e.\u003c/p\u003e\n\u003cpre tabindex=\"0\"\u003e\u003ccode class=\"language-mermaid\" data-lang=\"mermaid\"\u003egraph TD\n    A[Q1 2026 HR Strategy] --\u0026gt; B{Key Risks}\n    B --\u0026gt; C(Talent Risk)\n    B --\u0026gt; D(Operational Risk)\n    B --\u0026gt; E(Compliance)\n\n    C --\u0026gt; C1[Promotion Curse]\n    C1 --\u0026gt; C2[\u0026#34;17% Attrition \u0026lt;br/\u0026gt;(Recently Promoted)\u0026#34;]\n    style C2 fill:#ffcccc,stroke:#333,stroke-width:2px\n\n    D --\u0026gt; D1[Burnout Risk]\n    D1 --\u0026gt; D2[\u0026#34;39% Attrition \u0026lt;br/\u0026gt;(High Risk Segment)\u0026#34;]\n    style D2 fill:#ff9999,stroke:#333,stroke-width:2px\n\n    E --\u0026gt; E1[Diversity Health]\n    E1 --\u0026gt; E2[\u0026#34;Pay Gap \u0026lt; 2% \u0026lt;br/\u0026gt;(Target Achieved)\u0026#34;]\n    style E2 fill:#ccffcc,stroke:#333,stroke-width:2px\n\u003c/code\u003e\u003c/pre\u003e\u003ch2 id=\"-critical-risks-identified\"\u003e📌 Critical Risks identified\u003c/h2\u003e\n\u003ch3 id=\"1-the-promotion-curse-talent-risk\"\u003e1. The \u0026ldquo;Promotion Curse\u0026rdquo; (Talent Risk)\u003c/h3\u003e\n\u003cp\u003eWe traditionally view promotions as a retention tool. Our data proves the opposite.\u003c/p\u003e","title":"Q1 2026 Board Update: Strategic HR Insights"},{"content":"\nEvolving HR for the Era of AI Agents. The Context Workforce developments show a clear trajectory: the systems managing people are becoming as complex as the people themselves. Traditional administrative methods are no longer sufficient to support a modern, distributed organization that is evolving rapidly. As AI agents become a fundamental part of operations, the very definition of \u0026ldquo;labor\u0026rdquo; is changing. HR must evolve with it.\nThe Perspective We are moving away from viewing employees as static resources and toward viewing the organization as a dynamic ecosystem. My approach is not about AI taking over; it is about using technology to improve the workforce. This evolution offers the potential to significantly boost productivity and return ownership of work to the people doing it.\nMy Role I operate as a Strategic HR Tech Partner. I understand the \u0026ldquo;People\u0026rdquo; side of the business, but I also master the \u0026ldquo;AI\u0026rdquo; tools required to build modern solutions. I bridge the gap between strategy and execution, using AI to prototype tools that make work more human, not less.\nHow I Build Strategy over Syntax. I do not write code from scratch; I architect solutions. I utilize AI agents to handle the technical execution (Python, SQL), allowing me to focus entirely on business logic, data validity, and rapid deployment.\nOrchestration: Google Studio AI, Google Antigravity, Gemini Pro \u0026amp; Agentic Workflows. Execution: Python \u0026amp; SQL (AI-Generated \u0026amp; Audited). Visualization: Web Applications, Interactive Dashboards, Power BI, Excel. Collaboration: Slack / Microsoft Teams, Linear / Jira, Miro, Notion. LinkedIn | GitHub\n","permalink":"https://0xthijs.github.io/about/","summary":"\u003cp\u003e\u003cimg alt=\"Strategic HR AI Banner\" loading=\"lazy\" src=\"/images/home-banner.png\"\u003e\u003c/p\u003e\n\u003ch1 id=\"evolving-hr-for-the-era-of-ai-agents\"\u003eEvolving HR for the Era of AI Agents.\u003c/h1\u003e\n\u003ch3 id=\"the-context\"\u003eThe Context\u003c/h3\u003e\n\u003cp\u003eWorkforce developments show a clear trajectory: the systems managing people are becoming as complex as the people themselves. Traditional administrative methods are no longer sufficient to support a modern, distributed organization that is evolving rapidly. As AI agents become a fundamental part of operations, the very definition of \u0026ldquo;labor\u0026rdquo; is changing. HR must evolve with it.\u003c/p\u003e","title":"About Me"}]