HR.Project archive

Selected project archive · 11 entries

Work you can watch, inspect, and share.

Applied AI, digital twins, immersive systems, and research prototypes—each documented with a focused walkthrough and a permanent link that opens directly at the project.

Explore the work

01 / Applied systems

Built for real use.

Products and prototypes spanning private AI, operational digital twins, industrial training, education, and browser-based 3D.

01Local AI · Agentic systems

Permanent link

Local AI Agent

A private coding agent that works across real repositories without sending project data to the cloud.

Project walkthrough02:00 · Silent screen recording

A local-first AI workspace that can map a codebase, gather context, read files, and carry out multi-step tasks through Ask, Edit, and Agent modes. The interface keeps the repository tree, live reasoning, actions, and results visible so the operator can follow what the system is doing.

What it demonstrates
  • Runs local models while keeping prompts and source files on the machine.
  • Combines repository-aware context, reusable skills, and self-hosted web search.
  • Requires explicit approval before potentially dangerous system actions.
Local LLMAI agentsRepository intelligenceSafety controls

02Digital twins · Operational intelligence

Permanent link

Digital Twin Dashboard

A spatial operations dashboard that connects a data-centre model with equipment and environmental intelligence.

Project walkthrough02:54 · Silent screen recording

An interactive digital-twin interface for moving from a whole-facility view to individual pods and assets. The 3D environment is paired with temperature, humidity, load, alarm, and trend panels, giving operational data the physical context needed for faster monitoring and investigation.

What it demonstrates
  • Navigable 3D facility and equipment-level views in the browser.
  • Linked telemetry, trends, gauges, and alarm-priority information.
  • A unified interface for spatial inspection and operational decision-making.
Digital twinReal-time 3DIoT telemetryData visualisation

03VR learning · Cable-jointing training

Permanent link

VRJ

A comprehensive VR cable-jointing training platform for students and early-career professionals.

Project walkthrough05:46 · Silent screen recording

VRJ is an immersive virtual reality (VR) training platform designed to teach cable jointing to students and early-career professionals. It provides a large library of VR training modules covering multiple cable-jointing types and complete procedures. Through guided, hands-on simulations, learners practise tool selection, cable preparation, measurements, component handling, assembly sequences, and safety-conscious techniques in a realistic and repeatable virtual environment before working with physical cables.

What it demonstrates
  • Developed by Hari Ragavan, with an existing companion dashboard for progress, results, feedback and reports.
  • A large training library spanning multiple cable-jointing types and end-to-end procedures.
  • Designed to help students and early-career professionals develop practical confidence.
  • Guided tool use, measurements, component handling, and procedural sequencing in VR.
Virtual realityCable jointingTechnical educationUnitySimulation training

04XR learning · Medical education

Permanent link

XR Human Anatomy

An immersive learning experience that makes the three-dimensional structure of the human body explorable.

Project walkthrough03:08 · Narration and music

Designed for students who need more than a flat textbook view, the experience lets learners isolate anatomical regions and systems, inspect structures such as the humerus, reveal landmarks and muscle attachments, move naturally around the body, and test what they have learned.

What it demonstrates
  • Region, skeletal-system, bone, and muscle-level exploration.
  • Direct selection reveals labels, landmarks, insertions, and origins.
  • Immersive inspection and self-assessment support active learning.
Extended realityHuman anatomyEducationInteractive 3D

05Web 3D · Design visualisation

Permanent link

Interior Visualization — Browser Based

A real-time interior walkthrough that brings design review directly into the browser.

Project walkthrough01:53 · Silent screen recording

A lightweight WebGL experience for navigating a furnished interior at human scale and evaluating design alternatives in context. The recording demonstrates free movement through the room and rapid changes to wall colours and finishes without leaving the live 3D scene.

What it demonstrates
  • Plugin-free browser delivery for immediate access and sharing.
  • Real-time navigation, lighting, materials, furniture, and spatial context.
  • Fast visual comparison of colour and finish options during design review.
WebGLReal-time 3DInterior designBrowser experience

02 / Research experiments

Questions made runnable.

Instrumented systems that turn research questions into software that can be used, measured, and evaluated.

R1Interactive data systems · HCI

Permanent link

Progressive Query Explorer

Does showing an approximate answer now help people explore more effectively than making them wait for an exact answer?

Project walkthrough00:29 · Silent screen recording

An instrumented, browser-native research prototype for studying how response latency and progressive refinement affect exploratory analysis. Users cross-filter one million property transactions across five linked views while switching between immediate sample-based feedback and conventional exact queries.

What it demonstrates
  • Exact results versus progressive sample-to-exact refinement.
  • One million records queried locally through DuckDB-WASM and Parquet.
  • Exportable timings, interactions, and exploration traces for evaluation.
Progressive computationDuckDB-WASMHCIParquet
Open the working prototype

R2Program synthesis · Human–AI interaction

Permanent link

Query by Example

Can a system infer the data transformation a person intends from only the result they demonstrate?

Project walkthrough00:58 · Silent screen recording

A deterministic symbolic program synthesiser that turns small input–output examples into ranked, executable query candidates. It searches a compact transformation language in a worker, keeps every program consistent with the demonstration, and exposes ambiguity rather than hiding it.

What it demonstrates
  • Searches filters, joins, groups, strings, sorting, and projections.
  • Makes every candidate inspectable as a structured program and generated SQL.
  • Benchmarks intent recovery on full tables the examples never reveal.
Program synthesisSymbolic searchWeb WorkersTypeScript
Open the working prototype

R3Business analytics · Customer decision systems

Permanent link

Retail Growth & Customer Intelligence

How do customer definitions and financial assumptions change a revenue-growth decision?

Project walkthrough00:57 · Silent screen recording

An interactive business analytics research prototype built from real historical retail transactions. It connects revenue exploration, behaviour-based customer segments, cohort retention and a transparent campaign simulator, allowing users to inspect the evidence and test how assumptions change the decision.

What it demonstrates
  • Research dataset: 506,073 cleaned merchandise transaction lines from UCI Online Retail
  • Decision workspace: Linked market filters, customer definitions, cohort evidence and campaign scenarios
  • Reproducible evidence: Explicit metric definitions, source fingerprints and exportable interaction traces
Business analyticsCustomer cohortsScenario analysisPythonTypeScript
Open the working prototype

R4Business analytics · Energy decision systems

Permanent link

Building Energy & Investment Planner

How do data coverage, forecasting baselines and tariff assumptions influence an energy investment decision?

Project walkthrough00:59 · Silent screen recording

A working research prototype connecting historical building electricity and weather data to operational and financial decisions. Users compare buildings, inspect overnight patterns and unusual days, evaluate a weather-informed model against a seasonal baseline, and test investment scenarios with every financial assumption exposed.

What it demonstrates
  • Research dataset: A reproducible subset of 17 building electricity meters from Building Data Genome 2
  • Measured comparison: Chronological holdout evaluation against a previous-week baseline on identical dates
  • Decision transparency: Visible observation coverage, tariff assumptions, savings sensitivity and exportable evidence
Business analyticsEnergy intelligenceModel evaluationInvestment scenariosPython
Open the working prototype

R5Immersive learning · Adaptive guidance

Permanent link

Adaptive Guidance in Industrial VR Training

Does reducing guidance as a learner succeeds improve later independent performance?

Existing VRJ platform · not the adaptive extension05:46 · Silent screen recording

A research extension grounded in VRJ, the industrial VR cable-jointing platform developed by Hari Ragavan. VRJ already includes visual and voice instructions, assessment modules and a companion progress dashboard. This browser prototype makes a proposed fading policy inspectable and logs help, errors and successful actions; the new policy has not been integrated into VRJ or evaluated with learners.

What it demonstrates
  • Existing system: Hari-developed VRJ with procedural interaction, audio guidance, assessments and dashboard reporting
  • Working extension: Fixed versus adaptive guidance policy, independent-mode demonstration and exportable browser event trace
  • Evidence boundary: Existing VRJ footage is implementation evidence, not a recording of the proposed adaptive condition
VRJUnity / C#Guidance fadingResearch instrumentation
Open the working prototype

R6Learning analytics · Instructor awareness

Permanent link

Instructor Awareness in Virtual Classrooms

Can an interpretable learner-progress dashboard help an instructor notice and respond to difficulties?

Instructor dashboard walkthrough · synthetic session00:48 · Silent screen recording

A working instructor-dashboard research prototype using an explicitly synthetic, replayable classroom session. It combines task progress, repeated errors and learner-initiated help requests into an inspectable support queue. Instructors can acknowledge cues, record their intended support action and export the trace. This is a browser prototype, not footage of a connected multi-user VR classroom or evidence of improved learning.

What it demonstrates
  • Research interface: Dashboard and baseline views preserve explicit help while varying analytical support
  • Observable events: Deterministic task-event replay with visible cue rules and recorded instructor acknowledgements
  • Responsible interpretation: No attention scores, headset-movement proxies, real learner records or invented study outcomes
Instructor dashboardEvent replayHuman–computer interactionTypeScript
Open the working prototype