A PERSONAL COLLECTIONAI SYSTEMS / SIMULATION / DIGITAL PRODUCTS

THE PERSON BEHIND THE SYSTEMS

PrabhasBangarugari.

AI systems. Simulation. Digital products.

SELECTED WORK / 2026
PRABHAS BANGARUGARI / PORTFOLIOPERSONAL PORTFOLIO / 2026

AI SYSTEMS SIMULATION DIGITAL PRODUCTS

PrabhasBangarugari.

01 / AN INTRODUCTION

I make complex ideas
visible and usable.

Autonomous agentic intelligence. Industrial digital twins. Cited retrieval. Praj Chess Engine. Interfaces built to be used.
Explore selected work
SELECTED WORK / AN INDEX01 — 04
01 /SteelSim + ACAMISSIMULATION + AGENTIC INTELLIGENCE 02 /RAG-Based AI AssistantPREVIEW THE RETRIEVAL SYSTEM 03 /Praj Chess EngineSOURCE + ENGINE NOTES 04 /Fashion CompanyLIVE STOREFRONT
FOUR DIFFERENT PROBLEMS. ONE PORTFOLIO.

01 / SELECTED WORK

Systems, made
tangible.

A plant you can run. An answer you can trace. An engine you can challenge. A storefront you can explore.

Illustrative steelworks with a glowing induction furnace 01 / INTERFACE IN CONTEXT BUILD SIMULATE REVIEW
AGENTIC INTELLIGENCE / INDUSTRIAL DIGITAL TWIN

SteelSim + ACAMIS

A steelworks.
A world to reason about. An industrial digital twin connecting a visual plant builder, deterministic simulation, and live telemetry. ACAMIS brings evidence-led monitoring and advisory reasoning into that simulated world. React FlowDeterministic simulationAgentic intelligence LIVE SIMULATION · EVOLVING INTELLIGENCE OPEN LIVE SIMULATION
BEHIND THE BUILD / SteelSim + ACAMIS

THE SYSTEM

I built a plant-modelling and simulation environment that makes equipment, process flow, and operating state inspectable.

THE ENGINEERING

Authoritative backend state reaches the interface through WebSocket telemetry. Invalid topology is gated before a run starts. Recorded frames make incidents replayable.

THE FRONTIER

Signal review and optional model advice are implemented. Broader autonomous optimization remains a direction for ACAMIS; consequential recovery follows approval gates.

INSPECT THE REAL INTERFACE SOURCE & TECHNICAL NOTES
Concept illustration of a connected knowledge archive 02 / CONCEPT STUDY QUESTION RETRIEVE CITE
RETRIEVAL / SOURCE ATTRIBUTION

RAG-Based AI Assistant

Every answer needs
somewhere to point. An offline handbook assistant built around BM25-style retrieval and page-level citations. A question becomes a path back to the material it came from. Offline retrievalBM25-style rankingCitations LOCAL SYSTEM · PUBLIC PREVIEW AHEAD EXPLORE THE PREVIEW
BEHIND THE BUILD / RAG-Based AI Assistant

THE QUESTION

How can someone find useful guidance in local handbooks and check the source behind it?

THE APPROACH

Relevant passages are ranked locally and paired with page references. The implemented retrieval layer is the foundation of this assistant.

CURRENT STATE

The local retrieval experience exists. A public interface is in preparation; a complete generative RAG workflow is not claimed here.

Concept illustration of a polished chess knight 03 / CONCEPT STUDY POSITION MCTS BEST MOVE
ENGINE DEVELOPMENT / MONTE CARLO TREE SEARCH

Praj Chess Engine

A move is the end
of a search. A Python UCI engine combining Monte Carlo Tree Search with Stockfish policy and value advice. Engine development extends from move selection to reliable protocol behavior. PythonUCI protocolMCTS PUBLIC SOURCE · UCI ENGINE EXPLORE THE ENGINE
BEHIND THE BUILD / Praj Chess Engine

THE ENGINE

I developed Praj to explore tree search and evaluation inside a chess engine that can communicate with a UCI-compatible interface.

THE HARD PART

Stopping and superseding a search must never leak an old move. The published audit covers cancellation, time controls, promotions, and bounded search memory.

THE EVIDENCE

Source, regression tests, and diagnostic results are available. Absolute Elo is unmeasured, and trained model weights are not published.

READ THE ENGINE AUDIT
Illustrative fashion showroom in warm editorial lighting 04 / INTERFACE IN CONTEXT DISCOVER BROWSE EXPLORE
INTERFACE DESIGN / DIGITAL COMMERCE

Fashion Company

Commerce, with
a point of view. A storefront shaped around editorial presentation, category browsing, and responsive product discovery. A different domain, with the same attention to how an interface feels. Responsive interfaceProduct discoveryEditorial design LIVE STOREFRONT VISIT THE STOREFRONT
BEHIND THE BUILD / Fashion Company

THE EXPERIENCE

I built Fashion Company as a visual commerce experience, connecting an editorial first impression with browsable products.

THE DESIGN PROBLEM

Imagery, hierarchy, and category navigation need to work together across screen sizes. Product discovery should stay clear on a phone.

EXPLORE IT

The public interface is available to browse. Payment processing, order fulfilment, and production commerce infrastructure are not asserted by this portfolio.

INSPECT THE REAL INTERFACE

Built by Purna Sai & Prabhas Bangarugari.

02 / HOW I THINK ABOUT BUILDING

The interesting part is
making it work.

01 /

Model the reality.

SteelSim models equipment and process flow. Its interface makes those relationships legible before a run begins.

EXPLORE THE CONNECTION

A plant’s topology, material flow, and operating state belong in the same view. SteelSim connects the model to a run you can inspect.

SEE STEELSIM + ACAMIS
02 /

Show the reasoning.

The RAG assistant cites its sources. ACAMIS is designed to trace agent assessments, escalation, and human approval back to operating evidence.

EXPLORE THE CONNECTION

A retrieval result becomes useful when its evidence is visible. The assistant keeps source pages close to the answer so the reader can check them.

EXPLORE CITED RETRIEVAL
03 /

Make it usable.

Fashion Company pairs editorial presentation with product discovery. The same attention to hierarchy helps technical tools feel clear.

EXPLORE THE CONNECTION

Hierarchy, clear navigation, and considered feedback shape how someone discovers the product. The interface is part of the engineering.

SEE FASHION COMPANY

03 / THE PERSON BEHIND THE WORK

PB / 2026

Curiosity,
made concrete.

01 AGENTIC SYSTEMS02 INDUSTRIAL SIMULATION03 DIGITAL INTERFACES

I’m drawn to the moment an abstract idea becomes something you can inspect, question, and use.

I’m Prabhas Bangarugari, a computer science student focused on AI and machine learning. My work brings together autonomous agentic intelligence, industrial simulation, retrieval systems, and digital interfaces. SteelSim + ACAMIS connects a plant’s simulated state with evidence, monitoring, and reviewable decisions.

I’ve also built an offline handbook assistant, developed Praj Chess Engine, and created Fashion Company. Each asks a different question. The discipline carries across: understand the system, expose its logic, and make the result useful.

PRABHAS / CURRENT PRACTICESIMULATION  ·  EVIDENCE  ·  INTERFACE
NOW /

AI systems and industrial simulation.

SOON /

Small Language Model — architecture, tokenizer, and training pipeline.

 FUTURE /

Business Operator — a future direction for evidence-led operations.

04 / A CONVERSATION

The next idea
starts with a hello.

A technical question, a collaboration, or a project with a little ambition. I’d like to hear it.

Start a conversation on LinkedIn
PRABHAS BANGARUGARI © 2026
GITHUB ↗ LINKEDIN ↗ BACK TO TOP ↑

Project interface

Actual project screenshot. Use Escape or Close to return to the portfolio.