|CHIRAG OS / AI VENTURE OPERATING SYSTEM

I engineer intelligent systems and build ventures around real-world problems.

AI engineer, researcher, and entrepreneur working across agentic AI, RAG, applied machine learning, explainable systems, multimodal AI, and AI-native product development.

CURRENTLY BUILDING: Cellvara - responsible AI adoption and governance infrastructure for pharma and biotech teams.

Agentic AIRAGApplied MLExplainable AIMultimodal SystemsAI Product StrategyStartup Building

COMMAND CENTER

Mission Control

A live overview of current ventures, engineering focus, professional context, and active research direction.

ACTIVE VENTURE

Cellvara

AI enablement and governance infrastructure for pharma and biotech SMEs.

Active VentureCustomer DiscoveryPrototype DevelopmentPilot Exploration
Explore Cellvara
ENGINEERING FOCUS

Current Engineering Focus

  • Agent orchestration
  • Enterprise RAG
  • Structured outputs
  • Human-in-the-loop workflows
  • AI evaluation
  • Explainable implementation guidance
  • Responsible AI adoption
PROFESSIONAL ROLEPending Verification

Hilti

Working Student AI

Working on Agentic AI-related initiatives in an industrial context, combining agentic AI, data, workflow thinking,embedded systems knowledge and practical business use cases.

Details pending verification

EDUCATION

Neu-Ulm University of Applied Sciences

M.Sc. International Entrepreneurship, Digitalization and Sustainability

EntrepreneurshipDigital transformationBusiness modelsSustainabilityImpact measurement and analysisVenture development
LATEST PUBLIC BUILD

PrepAI

AI interview-preparation platform that converts a résumé and job description into a personalised mock interview and coach-style report.

LangGraphFastAPIGradioLangChain
RESEARCH QUESTION

How can organisations adopt advanced AI systems without sacrificing explainability, governance, operational trust, or human accountability?

Responsible AIGovernance

ACTIVE VENTURE

Current Mission: Cellvara

Responsible AI adoption infrastructure for pharma and biotech teams.

The Problem

Fragmented AI tooling across departments
Unclear implementation pathways for regulated workflows
Compliance concerns without traceable evidence
Limited internal AI expertise in SME teams
Expensive and inaccessible external consulting
Difficulty matching AI tools to real workflows
Lack of traceable implementation evidence

AI Consultant-Grade Recommendations

  • Explainable implementation steps
  • Evidence-backed recommendations
  • Traceable reasoning
  • Action-oriented workflow guidance
  • Tool and process suggestions mapped to business goals

Role-Specific Compliance-Aware Enablement

R&DQuality assuranceRegulatory affairsPharmacovigilanceOperationsLeadership / innovation teams

Workflow Mapping and Tool Orchestration

  • Map workflows to appropriate AI tools
  • Evaluate tool fit and readiness
  • Connect tools into coherent operating workflows
  • Reduce disconnected AI experimentation
  • Support human review and governance
Active VentureCustomer DiscoveryPrototype DevelopmentPilot ExplorationPrivate Repository

PUBLIC BUILDS

Featured AI Systems

A selection of public AI systems and prototypes exploring agent orchestration, retrieval, multimodal intelligence, structured outputs, and workflow automation.

DocChat

Verified multi-agent RAG for reliable document intelligence.

Reliable RAG and verification architecture

RAGLangChainVector DBVerificationDocument AI

PrepAI

AI interview-preparation platform that converts a résumé and job description into a personalised mock interview and coach-style report.

Orchestration, structured outputs, APIs, schemas, and evaluation

LangGraphFastAPIGradioLangChainWhisper

MedAssist MAS

Safety-oriented multi-agent educational triage assistant with structured outputs and deterministic red-flag checks.

Safety-aware agents and structured generation

AutoGenGradioPydanticHealthcare AIMulti-Agent

Educational decision-support demo only. Not a diagnostic or medical-care system.

Speech Analyzer AI

AI meeting assistant that transcribes spoken content, normalises domain terminology, and creates structured minutes and action items.

Speech AI, domain adaptation, and workflow automation

WhisperSpeech AILangChainIBM watsonxMeeting Automation

AI NourishBot

Multimodal nutrition assistant that interprets food images and generates ingredient, nutritional, and recipe insights.

Multimodal AI and coordinated specialist agents

Multimodal AIVisionCrewAIWatsonXNutrition AI

Banking Fraud Detector

Production-oriented fraud-detection architecture designed around latency, recall, throughput, drift, and security constraints.

Production requirements and measurable system design

Applied MLMLOpsKafkaDrift DetectionSecurity

Vidintel

YouTube intelligence application for transcript extraction, summarisation, and grounded video question answering.

Video intelligence and retrieval pipelines

YouTubeFAISSWatsonXRAGSummarisation

Job Applying Buddy

CrewAI-based workflow that generates tailored job-application materials from a vacancy, résumé, GitHub profile, and personal summary.

Document automation and coordinated role-based agents

CrewAIGradioResume AICover Letter AIPDF Generation

BEYOND CODE

Venture Portfolio

Beyond building models and applications, I explore how AI can become useful products, validated workflows, and scalable ventures.

Cellvara

Customer Discovery / Prototype Development

Cellvara helps regulated pharma and biotech teams identify, evaluate, and implement AI tools and workflows with explainable guidance, role-specific recommendations, and audit-ready reasoning.

Responsible AI adoptionWorkflow mappingCompliance-aware guidanceTool recommendationsAI implementation planning

ForgeFoundry

Venture Experiment

ForgeFoundry automates startup strategy with multi-agent pipelines covering problem-solution framing, market research, ICP personas, business models, GTM strategy, landing page publishing, and AI-generated pitch decks.

Problem-solution framingMarket researchICP and personasBusiness modelGTM strategyAI-generated pitch deck

Afterz

Paused

Afterz was a prototype concept exploring real-time event discovery, social coordination, and safety-aware nightlife planning. Current status needs verification.

Event discoverySocial coordinationSafety and inclusion

SYSTEMS THINKING

Architecture Explorer

Explore how these AI systems are designed: from user input to agents, tools, retrieval, validation, and final outputs.

Click a node to explore

MEASURABLE OUTCOMES

Proof & Impact

Defensible evidence of engineering depth, research contributions, and venture building.

50+Public Repositoriesgithub.com/PRONGS-CHIRAG
PatentHolderuser-provided
PublishedResearcheruser-provided
8+Agentic AI & RAG Systemspublic GitHub
FounderCellvarauser-provided
IndustrialML Experienceuser-provided

Industrial Context — Bosch eBike Systems

Pending Verification

Industrial Context — Pending Verification

  • 100,000+ rows/day industrial data workflows
  • ETL pipeline: 3 hours → 8 minutes
  • Applied ML & explainability for e-bike acoustic/vibration data

FOUR TRACKS

Braided Journey

From computer science foundations through research, industrial ML, agentic AI, and venture building.

EngineeringResearchEntrepreneurshipEducation
engineering
engineering

Hilti

Working Student Agentic AI Engineer

May 2026 — Present

engineering

Bosch eBike Systems

Data Scientist

Jun 2023 — Mar 2025

research
research

FORWISS, University of Passau

Research Assistant

Jun 2022 — Jul 2023

SPANS ENGINEERING + RESEARCH
researchPending Verification

Bosch eBike Systems

[Pending Verification]

entrepreneurship
entrepreneurship

Cellvara/Zestlearn

Founder / Builder

[Pending Verification]

entrepreneurship

Cellvara

Founder

Jan 2024 — Present

education
education

PES University

[Pending Verification]

education

University of Passau

[Pending Verification]

education

Neu-Ulm University of Applied Sciences

[Pending Verification]

KNOWLEDGE WORK

Research & Innovation

Academic research, industrial applied ML, and intellectual property contributions.

Predicting Perceived E-Bike Motor Loudness and Sound Quality ratings using Vibration Data with Explainable Artificial Intelligence

Master's Thesis — Bosch eBike Systems

Predicting perceived e-bike motor loudness and sound quality ratings using vibration data with explainable artificial intelligence techniques.

Explainable AIApplied MLSignal ProcessingIndustrial AIEBM
Pending Verification

Wavelet Signal Compression Research

Research Assistant — FORWISS, University of Passau

Signal-processing research on compression of wavelet-based sensor data using NetCDF and HDF5 formats and JPEG compression techniques.

Signal ProcessingWaveletsNetCDFHDF5Data Compression

Smart Infrastructure and Connected Living Spaces

IoT and Embedded Systems Research

Research into smart infrastructure, connected living spaces, sensors, and embedded systems for intelligent environments.

IoTSmart InfrastructureEmbedded SystemsSensors

A Flying Object for Dispensing Chemical Composition and Method Thereof

Patent

Patent for an autonomous drone system designed for precision dispensing of pesticides in farms using computer vision algorithms.

Patent No: 201941047691
PatentDroneAgricultureAutomation Computer Vision

Research Interests

Agentic AIAutonomous AI systemsExplainable AIHuman-AI collaborationAI for regulated environmentsAI governance and enablementSafe and Responsible AI

SKILLS WITH PROOF

Capability Map

Every skill linked to a system that uses it. Not a logo wall.

LangGraph
PrepAI
CrewAI
Job Applying BuddyAI NourishBot
AutoGen
MedAssist MAS
Specialist-Agent Workflows
DocChatForgeFoundry

VERIFIED LEARNING

Credentials

Certifications and courses across agentic AI, deep learning, multimodal AI, and product development.

IBM

Building AI Agents and Agentic AI Workflows Specialisation

Agentic AI & RAGVerify

IBM

Vector Databases for RAG: An Introduction

Agentic AI & RAGVerify

IBM

Advanced RAG with Vector Databases and Retrievers

Agentic AI & RAGVerify

IBM

Agentic AI with LangGraph, CrewAI, AutoGen and BeeAI

Agentic AI & RAGVerify

IBM

Agentic AI with LangChain and LangGraph

Agentic AI & RAGVerify

IBM

Fundamentals of Building AI Agents

Agentic AI & RAGVerify

Anthropic

Introduction to Model Context Protocol

Agentic AI & RAGVerify

Anthropic

Model Context Protocol: Advanced Topics

Agentic AI & RAGVerify

WRITING & SHARING

Public Thinking

Sharing ideas on AI systems, building Cellvara, venture lessons, and applied research.

Agentic AI Systems

Designing Multi-Agent Systems That Don't Hallucinate

How structured outputs, verification agents, and deterministic pre-checks reduce hallucination in agentic AI systems.

Pending Verification
RAG & Retrieval

Why RAG Alone Isn't Enough

Retrieval-augmented generation improves grounding but verification agents are needed for production reliability.

Pending Verification
Building Cellvara

What I Learned from Customer Discovery in Pharma AI

Observations from speaking with pharma and biotech teams about AI adoption challenges and decision-making.

Pending Verification
Responsible AI

Responsible AI in Regulated Environments

Why explainability, traceability, and human oversight matter more than model accuracy in regulated sectors.

Pending Verification

THE PERSON

About Chirag

I design and build AI systems across agentic AI, RAG, applied machine learning, explainability, multimodal AI, and AI-native product development. My work connects technical depth with entrepreneurship, especially around responsible AI adoption in complex and regulated environments.

Photography

Capturing places and moments

Travel

Learning through new environments

Creative Writing

Poetry and reflective writing

Music & Creative AI

Exploring generative creativity

COLLABORATION PATHS

Work With Chirag

Whether you are hiring for AI roles, exploring a Cellvara pilot, building a startup, or looking for a research collaborator, choose the path that fits best.

Hire Chirag

Looking for an AI engineer who can ship agentic AI systems, RAG pipelines, and applied ML in real products?

AI engineering rolesAgentic AIApplied MLRAGAI product engineering

Explore Cellvara

Pharma or biotech team thinking about AI adoption? Let's discuss a pilot or proof of concept.

Pharma SMEsBiotech SMEsR&D teamsRegulatory teamsAI adoption leads

Build Together

Co-founder, researcher, developer, or designer interested in building AI products together?

Co-foundersResearchersDevelopersHackathon teamsStartup collaborators

Invite Chirag

Looking for a speaker, panelist, or workshop lead on AI engineering, agentic systems, or responsible AI?

TalksPanelsWorkshopsUniversity eventsStartup programmes

Research Collaboration

Working on explainable AI, agentic systems, regulated AI, or human-AI workflows?

Explainable AIAgentic systemsRegulated AIHuman-AI workflows