AI Architect · Conversation Designer · Technical Enabler

Turning GenAI hype into lean, deterministic, production-grade architectures.

A prototype is easy; a cost-effective AI system that scales without burning your cloud budget is engineering.

Over 9+ years, I’ve architected enterprise voice, chat, and multi-agent systems across 10+ global markets, building the developer tools that multiply engineering velocity and mentoring teams to ship beyond fragile API wrappers.

Senior Manager, Conversational AI Architect at VOIS · Since 2022

Book a consultation Open Caliper Open Lumen
+20%
Intent accuracy & tariff sales conversion lift via RAG pipelines
25–50%
Faster build & test cycles via reusable playbooks & CI/CD tools
10+
Markets: enterprise GenAI, safety guardrails & multi-agent systems shipped
1,000+
Hrs/yr of engineering time saved by custom automated testing engines

// The Architect & The Enabler

The problem with "vibe-coding" (why execs & CTOs call me): AI has given everyone the power to build a demo in an afternoon, but most of it is just stitching together bloated API calls without thinking about operational cost, context-burn, or production determinism. Because I spent nine years taking enterprise voice and chat systems from whiteboard to production, I know exactly where conversational and agentic systems fracture under real-world load.

The builder (lean & local-first engineering): I bridge human-centric conversation architecture and hardcore backend engineering. Whether I am designing an 11-agent MCP ecosystem for a global telco or writing air-gapped, offline-first semantic tools and local SLM routers in Python and FastAPI, I optimize for high ROI, deterministic control, and zero wasted tokens.

The enabler (multiplying team velocity): I don’t just build applications; I build the engine room that helps entire teams build faster. From browser-based prompt linters and topology mappers to enterprise Centres of Excellence and AI upskilling bootcamps, I turn teams of prompt-tinkerers into disciplined AI engineers.

// Core Capabilities

Pillar 01

Building the Right Toolsets

High ROI, low cloud burn

Designing production multi-agent systems, RAG intent pipelines, and local/air-gapped model architectures that deliver enterprise scale without bankrupting cloud budgets.

Multi-Agent SystemsMCP IntegrationRAG PipelinesLocal & Air-Gapped SLMsPythonFastAPIGoogle CCAI / Dialogflow ES
Pillar 02

Multiplying Engineering Productivity

The engine room

Building deterministic developer tools, automated LLM evaluation harnesses, and regression pipelines that strip friction from daily workflows and cut build/test cycles by up to 50%.

AST Prompt LintingMulti-Dataset LLM EvalsAutomated NLP TestingCI/CD Pipeline AcceleratorsNodeJSAngular
Pillar 03

Strategic AI Upskilling & Governance

The enabler

Establishing Centres of Excellence, cross-market delivery playbooks, and enterprise safety controls while mentoring teams to engineer deterministic AI workflows.

AI Safety & Red-TeamingPII & Jailbreak ControlsCoE LeadershipVoiceflow PrototypingRecipient Experience Design (RXD)

// The Non-AI Flagship

Deterministic governance, zero inference. Built by me, free to open in your browser.

The Caliper

// Regulate. Evaluate. Orchestrate.

> SYSTEM INTENT An integrated engineering suite designed to strip speculative hype from Generative AI. Enforce financial rigor, stress-test logic boundaries, and forge hardened assets for surviving use cases.

01

Crucible

Weigh. Burn. Decide.

Evaluate viability: a go/no-go score for an AI use case.

 02

The Matrix

Architect. Map. Bind.

Map a multi-agent design and dry-run every scenario, no AI needed.

 03

Forge

Construct. Format. Forge.

Construct the asset: structured, graded prompts.

 04

The Anvil

Stress. Fracture. Fortify.

Break the asset: red-teaming and quality runs.
Open Caliper

// The AI Flagship

Context-efficient AI execution. Built by me, running on my own machine. In private development.

Lumen

// The AI toolkit that boosts AI productivity.

> SYSTEM INTENT Small language models on your own machine do the specific jobs, so the frontier model only reads what matters. Offline, private and measured: your AI assistant gets 20× less context and answers in milliseconds.

01

CodeSearch

Ask your code. Get the exact lines.

Finds code by meaning or by name for Claude Code, Copilot or Cline, and shows what a change will touch.

 02

CodeGraph

See the logic. Navigate the build.

A live map of how your files connect, inside CodeSearch.
Open Lumen
Real run, real speed: CodeSearch answering a plain-English question.

// Experience

// Projects

All projects on The Craft

// Education & awards

Education

Bachelor of Technology

College of Engineering, Bhubaneswar

Information Technology

Aug 2012 – May 2016

Awards

Outstanding Performer

2017 – 2020

Bronze Pillar Award

2018 – 2019

Certifications

Google

• CCAI Customer Experience• Dialogflow ES

Cognigy

• Project Management for Conversational AI

Core Tech

• Python• Angular• NodeJS

// Writing

Codified blueprints: my architecture frameworks, publicly documented long before the current GenAI hype cycle.

Read all whitepapers →
Guide

Conversational AI architecture & conversation design

Voicebots, chatbots and IVR that work in production, and when to move from NLU to LLMs.

 Guide

Production GenAI: RAG, evaluation, safety & cost

What it takes to get generative AI from pilot to production in an enterprise.

 Guide

Agentic AI: multi-agent systems that survive production

Agent topology, MCP tools, evaluation and budgets, from an 11-agent build.

 IVR · Vol 1

Comprehending the Stumbling Blocks

Why most enterprise IVR implementations fail: the design traps, NLU pitfalls, and organisational blind spots.

 IVR · Vol 2

The Journey Towards Elegance

A framework for building IVR experiences that actually contain calls and earn user trust.

 IVR · Vol 3

Looking Beyond the Abyss

Where Conversational IVR goes next: Gen AI augmentation, RAG-based intent recognition, and the next generation of voice experiences.

 Whitepaper

Recipient Experience Design (RXD)

A conversation design system, with core principles for scalable architecture.

 Whitepaper

Telecom's Role in Disaster Relief & Resilience

How telecommunications infrastructure and conversational AI systems help in disaster response.

 Blog

Articles on Medium

Shorter pieces on conversation design and Gen AI.

// Built in the Open: The Engineering Ecosystem

Most architects only have slide decks. Here is my live engineering ecosystem, proving that you need both deterministic non-AI guardrails and context-efficient AI execution to ship reliable systems.

// Questions

Who is Shashikant Jha?

Shashikant Jha is a Conversational AI, GenAI and agentic AI architect based in Pune, India. He is Senior Manager, Conversational AI Architect at VOIS, with 9+ years building enterprise voicebots, chatbots, RAG pipelines and multi-agent systems across 10+ markets, and the builder of The Caliper and Lumen.

What conversational AI work has Shashikant Jha done?

He designed a globally deployed service-desk voicebot across IVR and chat, led Mindtree's Dialogflow Centre of Excellence, consulted on a US government chatbot, and replaced a legacy NLU with a RAG-based intent pipeline that improved intent accuracy by 20%. He also wrote a three-volume Conversational IVR whitepaper series. Conversational AI guide →

What agentic AI and GenAI experience does Shashikant Jha have?

He designed an 11-agent GenAI ecosystem with cross-agent communication, MCP integration and telemetry for a global telco, built LLM evaluation frameworks across multiple datasets and reasoning configurations, and set up jailbreak, PII and content-validation controls across 4+ markets. Agentic AI guide → · GenAI guide →

Does Shashikant Jha offer AI consulting?

Yes. He offers 15 to 60 minute consultations to audit enterprise GenAI architectures, design lean multi-agent workflows, and set up internal AI governance and developer testing pipelines. Book at cal.com/shashikant-jha-xthhjb/consultancy. He is also open to architect roles.

// Contact

Let's build something great

Whether you're designing a voice bot, building a chat experience, or exploring what Gen AI can do for your product, I'd love to hear from you.

For teams

Consulting

Dedicated sessions to audit enterprise GenAI architectures, design lean multi-agent workflows, or set up internal AI governance and developer testing pipelines. Choose a 15 to 60 minute slot.

Book a consultation
For hiring

Roles

My full résumé, experience and LinkedIn profile.

LinkedIn