RAJAT S
Let’s talk

AI/ML EngineerFull-Stack Developer

Open to collaborations & engineering roles

Hello, I'm RAJAT S

Rajat S in profile, chin resting on his hand, in a Microsoft AI/ML intern hoodie

I build intelligent systems.

Intelligent Systems • AI Agents • Product Engineering

I'm a computer science engineering student focused on AI/ML, full-stack development and intelligent automation. I build practical, end-to-end products, from AI-powered desktop experiences to evidence-driven compliance systems and browser automation.

View workContact

Delhi NCR, India

Scroll to discoverSelected systems / 2026

01 / Selected work

Work that thinks.

4 projects shown

An AI-powered desktop layer that brings voice, vision, planning and system control into one local assistant.

An evidence-backed compliance system that turns a photo of a product label into a verdict citing the exact legal clause.

An AI-assisted browser automation system that plans, executes and independently verifies multi-step web research.

A scam-shield that follows how a UPI, banking or wallet scam unfolds across messages, call notes, links and screenshots.

Built with

The tools I reach for.

  • GitHub
  • VS Code
  • Cursor
  • Claude Code
  • Docker
  • Ollama
  • FastAPI
  • Next.js

Also on GitHub

More experiments.

Engineering under the hood

From idea. To intelligent system.

One mission. Every layer of the system.

Drag to see OmniPilot's architecture become the working product.

OmniPilot / 2026

TypeScript · Playwright · Evidence

The OmniPilot results screen from its offline demo, labelled as fictional vendor data, showing sources captured and verified deliverables.
BlueprintLive build
Drag to see the systemOffline demo · fictional vendors

02 / Behind the systems

More than models.Engineering systems.

I'm pursuing a B.Tech in Computer Science and Engineering with an Applied Mathematics specialization at Vivekananda Institute of Professional Studies (VIPS), affiliated with Guru Gobind Singh Indraprastha University (IPU). In 2026 I joined Microsoft for an AI/ML engineering internship, from August to November, and was selected for the Google Student Ambassador Program. What I enjoy most is turning ideas into working products: from architecture and experiments through implementation, testing and deployment.

My work sits at the intersection of AI/ML, software engineering, automation and product development.

Rajat playing an acoustic guitar in a Microsoft office lobby
Between builds.

Expertise & skills

Education

B.Tech, Computer Science and Engineering

2025–2029

Applied Mathematics specialization · CGPA 9.1 / 10

Vivekananda Institute of Professional Studies (VIPS), Guru Gobind Singh Indraprastha University (IPU)

CBSE, Mathematics and Computer Science

2021–2024

Class 12: 90% · Class 10: 98.25%

Delhi Public School (DPS)

Capabilities by domain

  • AI & machine learning

    • AI / ML
    • LLM applications
    • OCR

    In: PraMaan · SafeCircle · Ultramatrix

  • Generative AI

    • LLMs
    • RAG
    • LangChain
    • NLP
    • Agentic AI

  • Agents & automation

    • AI agents
    • Browser automation

    In: OmniPilot · Ultramatrix

  • Full-stack

    • React
    • Next.js
    • Node.js
    • FastAPI
    • REST APIs

    In: All four systems

  • Languages & foundations

    • Python
    • TypeScript
    • JavaScript
    • C++
    • DSA
    • Docker
    • Git

Daily toolkit

  • GitHub
  • VS Code
  • Cursor
  • Claude Code
  • Docker
  • Ollama
  • FastAPI
  • Next.js

From problem to system

  1. Problem discovery

    Start from what actually breaks for a person: a label nobody can check, or a scam spread across several messages.

  2. Research & system design

    Read the primary sources and constraints, then decide where AI belongs and where it does not.

  3. Architecture & implementation

    Separate observation from decision, and build behind interfaces so a better model can drop in later.

  4. Testing & validation

    Test the failure paths, measure on held-out data, and publish the limits next to the numbers.

  5. Deployment & iteration

    Containerise, re-run from clean, and write down exactly what was and was not verified.

AI observes.Rules decide.Tests prove.

The principle behind PraMaan, and the standard I hold the rest of my work to.

03 / The journey

Keep building. Keep testing.

Experience & recognition

  • Noida · Hybrid

    Selected for an AI/ML engineering internship at Microsoft, working hybrid from Noida. The work itself is internal, so its details stay private.

04 / Your next system

Let's build together.

Got a hard problem or an engineering role in mind?

Let's make something that works in the real world.

Start a conversation
The SafeCircle Guardian attack chain on synthetic demo data: evidence items linked to stages from approach to extraction.Built to be
verified.Live build

SafeCircle Guardian: attack chain, demo data