Skip to content

Shravan P Jois

Shravan P Jois in a dark suit, white shirt and tie, smiling at the camera; black and white, on a cobalt panel.

I build agentic systems that survive production.

Bengaluru
1st prize, TietoEvry TechUtsav 2025

View work

About

Shravan P Jois in a grey suit and tie, smiling in an office corridor with colleagues talking behind him.

I build AI solutions, agentic systems and automation workflows, from proof of concept to deployment.

AI Engineer with 3 years building production-grade Generative AI and agentic systems for enterprise teams. At TietoEvry that means LangGraph multi-agent systems, an AI-SDLC framework that automates requirement analysis, code generation, review, testing and documentation, enterprise agents on Microsoft Copilot Studio and Microsoft 365 Copilot, a GenAI clinical notes summariser (RAG + LangChain) and an ML + LlamaIndex fraud anomaly engine.

Before that, at Plugeasy Technologies, I built the backends behind EV charging: Python/FastAPI microservices on OCPP 1.6 and OCPI, REST APIs for the apps and charge points, MongoDB design and database automation for live charging sessions, all running on AWS.

Award

1st Prize

TietoEvry TechUtsav 2025

For EVA (Digital Human), the AI-SDLC multi-agent framework, and enterprise GenAI capability demos presented to senior leadership.

Anyone can call a model. Engineering is making it answer right twice.

Quote by Shravan P Jois

The model is in every box. The system makes it reliable.

One request, traced

Illustrative trace: an IT helpdesk request traced through a supervisor, memory, a retriever, an action agent with a tool call, evals and guardrails, and the response. Timings and token counts are illustrative.
helpdesk-agent/ tr_9b2e41c7queuedrunningok
User requestChat

My laptop won't connect to the office VPN since this morning. I have a client call at 3 pm.

Latency
2.14 s
Tokens
2,324
Spans
8
  1. Supervisor2.14 s412queuedrunningok
  2. Memoryread84 ms–queuedrunningok
  3. Retriever412 ms186queuedrunningok
  4. Knowledge basesearch296 ms–queuedrunningok
  5. Action agent868 ms642queuedrunningok
  6. Tool callcreate_ticket641 ms–queuedrunningok
  7. Evals + guardrails233 ms211queuedrunningok
  8. Response491 ms873queuedrunningok
Supervisoragentsp_01
2.14 s412 tokensqueuedrunningok

Input

VPN won't connect since this morning; client call at 3 pm.

Output

Plan: read this user's memory, retrieve the VPN runbook, act through the ticketing tool, check the answer, then reply.

Memorymemorysp_02
84 msno model callqueuedrunningok

Input

user u_7f3a · last 30 days

Output

Six days ago: VPN certificate renewed on this laptop. Windows 11, VPN client 5.2.

Memory

Retrieverretrieversp_03
412 ms186 tokensqueuedrunningok

Input

Request + memory: recent certificate renewal, client 5.2.

Output

Query: VPN fails after certificate renewal, Windows client 5.2. Filter: network runbooks.

Context design

Knowledge basesearchsp_04
296 msno model callqueuedrunningok

Input

top_k 6 · filter network runbooks

Output

3 chunks: VPN runbook §4 Certificate re-enrolment, known issue KB-1129, client 5.2 release notes.

Action agentagentsp_05
868 ms642 tokensqueuedrunningok

Input

Runbook §4 + known issue KB-1129.

Output

Runbook §4 fits. Re-push the certificate and open a ticket so the network team can follow up.

Tool calltoolsp_06
641 msno model callqueuedrunningok

Input

create_ticket(queue: network, priority: P2, user: u_7f3a)

Output

200 OK · ticket HD-4821 created · certificate re-push queued for this device.

Clear tool contracts (MCP)

Evals + guardrailsevalsp_07
233 ms211 tokensqueuedrunningok

Input

Draft reply + sources + tool result.

Output

Grounded in 2 sources · no personal data in the reply · tool result matches its contract · pass.

Evals on every change

Responsellmsp_08
491 ms873 tokensqueuedrunningok

Input

Supervisor: send the checked reply.

Output

Your VPN certificate is being re-pushed now. Restart the VPN client in about 5 minutes. If it still fails, ticket HD-4821 is with the network team before your 3 pm call.

A supervisor that owns the outcome

  1. Cause: No memory of what happened last timeFix: Memory

  2. Cause: Missing or stale context at the moment the agent needs itFix: Context design

  3. Cause: Tool calls returning something nobody planned forFix: Clear tool contracts (MCP)

  4. Cause: No way to tell if a change made things better or worseFix: Evals on every change

  5. Cause: Two agents and no clear owner of the final answerFix: A supervisor that owns the outcome

How I work

Four short chapters, in my own words.
Shravan P Jois, smiling, in a navy suit, white shirt and burgundy tie; a colour photograph on a light studio backdrop.
A photograph with depth. It follows the pointer.

I'm Shravan, an AI Engineer in Bengaluru. I started at Verzeo in 2021 as a machine learning intern, and finished my B.E. in Electronics and Communication in 2023. At Plugeasy I built the backends behind EV charging. Since July 2025 I have been building agentic systems at TietoEvry.

I build agentic systems for enterprise teams. Multi-agent workflows on LangGraph, where a supervisor owns the final answer. Retrieval that grounds every answer in a team's own data. And enterprise copilots on Copilot Studio and Microsoft 365 Copilot.

Every system starts as a small proof of concept. Then come the agents, each with a clear tool contract and a supervisor that owns the outcome. Evals and guardrails run on every change, so I can tell whether it got better. Only then does it go to production.

Outside work, I design and build websites as a hobby. An ice cream brand, a hair atelier, a bilingual café in Leipzig, and a few more. This site is one of them, designed and built by me.

Selected systems

06 systemsAgentic, retrieval and backend work from enterprise and independent builds.
System 1 of 62025–present
RoleArchitect, TietoEvry

AI-SDLC Multi-Agent Framework

A LangGraph framework at TietoEvry where one supervisor agent coordinates specialised agents for requirement analysis, code generation, peer review, unit tests and documentation.

Stack

LangGraph, LangChain, Python, FastAPI, Azure OpenAI, GitHub Copilot, REST

Read the case
System 2 of 62025
RoleEngineer, TietoEvry

EVA, Digital Human

An emotional virtual assistant: a multimodal persona that combines NLP sentiment analysis, real-time computer vision, real-time speech synthesis and generative AI. It won 1st prize at TietoEvry TechUtsav 2025.

Stack

Python, PyTorch, Azure OpenAI, LangChain, Pipecat, Computer vision

Read the case
System 3 of 6
RoleArchitect and engineer

Agentic IT Helpdesk

The same problem solved twice: an enterprise helpdesk agent on Microsoft Copilot Studio, and an independent full-stack platform where four cooperating agents triage, resolve, escalate and read sentiment.

Stack

Copilot Studio, Power Automate, Azure OpenAI, FastAPI, SQLAlchemy, Celery, Redis, PostgreSQL, React, TypeScript, WebSockets, Claude API

Read the case
System 4 of 6
RoleArchitect and engineer

WhatsApp AI Assistant SaaS

A multi-tenant SaaS that sells Claude-powered WhatsApp assistants to small businesses. An operator onboards clients; each business owner gets a scoped dashboard for the bot, its knowledge and human handoff.

Stack

Node.js, Express, Claude API, WhatsApp Business Cloud API, PostgreSQL, Prisma, React, Tailwind CSS, Docker

Read the case
System 5 of 6
RoleEngineer (TietoEvry), then independent build

Clinical Notes Intelligence

A clinical documentation assistant for therapists. It started at TietoEvry as a LangChain RAG summariser, then became an independent build: upload a session, get a transcript, summaries, SOAP and BIRP notes, and grounded Q&A.

Stack

Python, FastAPI, LangChain, Azure OpenAI, ChromaDB, Streamlit, React, TypeScript, ffmpeg

Read the case
System 6 of 6
RoleBackend lead (Associate Engineer); web redesign

EV Charging Platform, Plugeasy

Core backend microservices for Plugeasy's EV charging platform, implementing OCPP 1.6 and OCPI for real-time charge point management, plus a full redesign of plugeasy.in with measured Core Web Vitals.

Stack

Python, FastAPI, OCPP 1.6, OCPI, MongoDB, AWS, Next.js, TypeScript, Tailwind CSS

Read the case

Web Craft

Web development is the hobby. A selection of sites I have designed and built.

Showing all 7 sites

Jeweller, Fürth, home page

Jeweller, Fürth

A static trilingual site (German, Arabic, English) for a jeweller in Fürth, led by a live Three.js ring you can spin and a stone selector that swaps the gem in real time.

React, Three.js, Motion, i18n

Burger house scroll story, home page

Burger house scroll story

A concept site where a burger rotates, bursts apart and is rebuilt layer by layer as you scroll, followed by a full-screen menu with an animated add-to-cart.

HTML, CSS, JavaScript, GSAP

Premium ice cream brand, home page

Premium ice cream brand

A concept site for a small-batch ice cream brand told through scroll-scrubbed video: the hero cone tumbles into the first flavour card and a sundae builds itself scoop by scoop.

HTML, CSS, JavaScript, GSAP

Aparna Gold Buyers, home page

Aparna Gold Buyers

A static multi-page site for a gold buyer in J.P. Nagar, Bengaluru, built with Astro 5 islands. Every lead action is a phone call, a WhatsApp message or an email.

Astro, React, three.js, React Three Fiber

Bilingual café, Leipzig, home page

Bilingual café, Leipzig

A German and English concept site for a café and cake shop in Leipzig, on Next.js 15 locale routing with a typed translation contract.

Next.js, React, TypeScript, Tailwind CSS

E-commerce platform, home page

E-commerce platform

A production e-commerce template on Next.js 16: storefront, cart, real Stripe checkout, accounts and admin, all driven from one config file and runnable end to end with no API keys.

Next.js, TypeScript, Tailwind CSS, Prisma

Hair atelier and day spa, home page

Hair atelier and day spa

A concept site for a hair atelier and day spa with a WebGL aura in the hero, drawn by a bubble shader I wrote by hand after the stock material measured about 630 ms per frame.

React, Vite, Tailwind CSS, Framer Motion

All web work

Experience

AI Engineer · Bengaluru, India

  1. TietoEvry

    Junior AI Engineer

    Jul 2025 – PresentBengaluru · Hybrid

    • Architected an AI-SDLC multi-agent framework on LangGraph: a supervisor agent coordinates specialised agents for requirement analysis, code generation, peer review, unit test creation and documentation, each with custom skills and tool calling.
    • Built enterprise AI agents on Microsoft Copilot Studio and M365 Copilot: an IT Helpdesk Agent (ticket triage, intent classification, routing to sub-agents, escalation; lower MTTR) and a Requirement Engineering Agent using Microsoft Graph API.
    • Built a GenAI clinical therapy notes summariser (LangChain, RAG, Streamlit), significantly reducing practitioner review time.
    • Developed a banking transaction anomaly detection engine (ML + LlamaIndex retrieval) integrated with enterprise fraud mitigation workflows.
    • Established prompt engineering standards, LLM evaluation protocols and best practices adopted across the AI engineering team.

    Tools: LangGraph, LangChain, LlamaIndex, CrewAI, Azure OpenAI, AWS Bedrock, Copilot Studio, M365 Copilot, Power Automate, Graph API, MCP, RAG, ChromaDB, Pinecone, Weaviate, Streamlit, Python, FastAPI

  2. Plugeasy Technologies

    Associate Engineer

    Apr 2024 – Jun 2025Remote

    • Led core backend microservices for EV charging (Python, FastAPI), implementing OCPP 1.6 and OCPI for real-time charge point management and interoperability.
    • Designed RESTful APIs connecting the frontend, mobile apps and IoT charge points.
    • MongoDB schema design, indexing, query optimisation and database automation for real-time charging session data.
    • Deployed and maintained the backend services on AWS (EC2, S3, Lambda).

    Tools: Python, FastAPI, OCPP, OCPI, REST APIs, Microservices, MongoDB, AWS, Git

  3. Verzeo

    AI & Python Engineering Intern

    Sep 2021 – Oct 2021Remote

    • Built foundational ML pipelines and computer vision workflows (image classification, feature extraction).
    • Developed Python backend services for web applications.

    Tools: Python, ML, Computer Vision

  4. Jawaharlal Nehru National College of Engineering (JNNCE)

    B.E. Electronics & Communication Engineering

    Dec 2020 – Aug 2023Shivamogga, Karnataka

    Education

Toolkit

Hover, focus or tap a tool to see the case studies that use it.

01Agents and models

  • Model Context Protocol (MCP)
  • Context engineering
  • LLM evaluation
  • CrewAI
  • Hugging Face Transformers
  • Scikit-learn

02Platforms, cloud and data

  • Git
  • GitHub
  • GitLab
  • LLMOps and Responsible AI
  • Pinecone
  • Weaviate

03Web

Certifications

Eight certifications from three issuers, the latest in Oct 2026.
Eight certifications from three issuers, the latest in Oct 2026.
CertificationIssuerDate
GitHub Copilot Fundamentals (Part 1 & 2)MicrosoftOct 2026
Claude Code in ActionAnthropicOct 2026
Claude Code 101AnthropicOct 2026
Claude Platform 101AnthropicOct 2026
Introduction to Claude CoworkAnthropicOct 2026
AI Fluency: Framework & FoundationsAnthropicOct 2026
Claude 101AnthropicOct 2026
Internship Program: Artificial IntelligenceCredential 725768430VerzeoJan 2022

Why hire me

Twenty seconds, no sound: the case for hiring me as an AI Engineer, in my own words and interfaces.

What the film says

  1. 0:00The hookAnyone can call a model.Engineering is making it answer right twice.
  2. 0:03What I doI build AI solutions, agentic systems and automation workflows, from proof of concept to deployment.AI-SDLC framework · Agentic IT helpdeskTwo interfaces from the portfolio, built at TietoEvry
  3. 0:07What I have shipped3 years building production-grade Generative AI and agentic systemsLangGraph multi-agent systems and an AI-SDLC framework at TietoEvryEnterprise agents on Microsoft Copilot Studio and Microsoft 365 CopilotA RAG clinical notes summariser and an ML + LlamaIndex fraud anomaly engine
  4. 0:11How I shipContextTool contracts (MCP)MemorySupervisorEvalsThe model is in every box.The system makes it reliable.
  5. 0:15Recognised1st prizeTietoEvry TechUtsav 2025Bengaluru, India · Open to AI Engineer, Generative AI Engineer and Agentic AI roles
  6. 0:18The askHire me.The enquiry form and the résumé are right below.

Open to AI Engineer, Generative AI Engineer and LLM / Agentic AI Engineer roles.

Start a hiring enquiryDownload résumé

Composed in HTML and rendered with HyperFrames. Footage: Pexels (see scripts/README.md).

Contact

Let’s build the system, not just the model.

WhatsApp

A greeting is ready to send

Start a chat

Message me

Phone

Mobile, India (IST)

Call

Bengaluru IST

Hiring

Hire me for your team

Open to AI Engineer, Generative AI Engineer and LLM / Agentic AI Engineer roles.

Download résuméDesigned version

What you are building, where it stands, and when you need it. At least 20 characters.

Goes straight to my inbox.