WhatsApp

Twenty-two years of things that shipped.

Surbhi Goyal, founder of Kinexus
Surbhi Goyal
Founder, Kinexus
On LinkedIn →

Agentic AI infrastructure, AWS data platforms, HR technology used across the Nordics, and citizen services for a state government. Different decades, same test: did the person at the counter actually use it?

See the work → Talk about yours
10,000
Women entrepreneurs served by WISE
1 → 150+
Team scaled, attrition under 2%
<28 days
Data platform to production
22 yrs
Building and running software
Platforms & products

Built for enterprises, measured in production.

Every number below is against where that system started — not a vendor benchmark. Where the work is publicly documented, the link goes to the source.

Enterprise platform

AWS-Native Data Lakehouse

An AWS-native data lakehouse that goes from nothing to production in under 28 days — because the configuration work most teams do by hand is done for them.

  • Deploys in <28 days, not quarters
  • 70% less configuration effort
  • 35% lower infrastructure cost
Enterprise data platform
AI solution

Paperless AI

A GenAI document assistant for the documents legacy OCR kept getting wrong — handwriting, bad scans, forms nobody standardized. It reads them, then explains what it read.

  • 50% faster document processing
  • 52% fewer errors than legacy OCR
  • Built on AWS Bedrock
Paperless AI →
HealthTech

Multimodal Clinical AI Ecosystem

A clinical ecosystem pairing intelligent patient intake with continuous vitals monitoring — built to stay responsive with many clinicians on it at once, not just in a demo.

  • 5–7 second responses under concurrent load
  • Multimodal: text, documents, vitals
  • Intake and monitoring in one stack
HealthTech engagement
Workforce AI

Field Workforce AI

Field data used to arrive as photos and phone calls and get retyped into a system days later. The system orchestrates it at the edge on a hub-and-spoke design, then feeds enterprise analytics directly.

  • Hub-and-spoke agent architecture
  • Intelligent field-data orchestration
  • Wired into enterprise analytics, not a silo
Kinexus engagement
HRTech / HCM

Core HR & Talent Management

The Nordic region's leading HR platform, scaled over fifteen years. The monolith that got it there stopped being able to carry it, so we moved it to microservices — carefully, while customers stayed live on it.

  • 50% performance gain post-migration
  • Reporting speed optimized 9×
  • SOC 2 Type II and ISO 27001 governance
Enterprise HR platform
E-governance

Citizen Services Portal

Digitizing counter services for a state government in Java/JSP, back when the counter was the only interface a citizen had. The software either worked for the clerk at the counter or it didn't work at all — a lesson that never left.

  • Statewide citizen service digitization
  • Java / JSP, built for low-bandwidth reality
  • Where the adoption-first habit started
State government engagement
Open source

The code is public. Go read it.

Working systems, not toy notebooks. If you want to know how we think about agent architecture before you hire us, this is the honest way to find out.

Multi-Agent Medical Assistant

A specialized LangGraph architecture for healthcare triage and diagnostic support — agents that know when to escalate rather than guess.

LangGraphPythonHealthcare
View repository →

Healthcare Appointment Booking

An autonomous system for the scheduling problems that are never one step — rescheduling, constraints, and clean handoffs between agents.

Multi-agentOrchestrationPython
View repository →

EU AI Act Compliance Engine

A governance tool that turns a regulatory framework into executable controls — so compliance is something the system enforces, not something a spreadsheet claims.

GovernanceRiskEU AI Act
View repository →
The journey

Where the habits came from.

Surbhi Goyal, founder of Kinexus
Surbhi Goyal
Founder, Kinexus
Full profile on LinkedIn →

Twenty-two years, and the same job the whole way through: make the software actually get used. I started writing automation for enterprise BPM suites, moved into scrum and performance engineering on a Nordic HR platform, then spent seven years running product operations for it — growing the India organization from one person to 150+ with attrition under 2%, and owning the SOC 2 and ISO 27001 work that comes with holding other people's employee data.

Most recently VP of Products, taking AI-native platforms from idea to enterprise deployment — a GenAI document assistant in production, and a data lakehouse teams stand up in under a month. That's where the conviction hardened: AI belongs only where it earns its place, and everywhere else a simple system wins.

Every role taught the same lesson from a different angle: adoption is the deliverable, not installation.
The toolkit

What's actually in the hands.

Not a keyword list for a recruiter. These are the things that have been used to ship something someone paid for.

AI & agentic architecture

AI/ML strategyGenAI on AWS BedrockLangChainLangGraph Multi-agent systemsRAG architectureLLM & agent orchestration n8n automationPerplexity / Comet

Product & leadership

Product roadmappingOrganizational scaling (1→150)P&L ownership Zero-touch opsn+2 velocity engineEnterprise data architectureAgile / Scrum

Cloud & data infrastructure

AWS nativeLambda & Step FunctionsCloudFormationData lakes ETL pipelinesDynamoDBAuroraQuickSightSQL / NoSQL

Core engineering base

PythonJava / J2EENodeJSReactJSKotlin Spring BootRESTful APIsAngular 8+RxJS Gradle (Kotlin DSL)WebSphereTomcatJasper Reports

Governance & security

SOC 2 Type IIISO 27001GDPRIndia DPDPCIS Critical Security Controls

Enterprise tools & BI

TableauPower BIJIRAConfluenceYouTrackProductBoardFigma
Start here

Your problem is probably not on this page.

That's fine — none of these started as a case study either. Tell us what isn't working, and we'll tell you honestly whether we can help, what we'd do first, and what we'd leave alone.

Start a conversation → See Alvora Hire