      Yankit Rajor
      AI Engineer · Software Developer
      Delhi NCR, India
      
        Open to conversations

Fig. 01 — Nine components, one system · click to reassemble

I turn complex workflows into intelligent software.

I'm an AI Engineer and Software Developer building intelligent agents, business automations, and full-stack systems that solve real operational problems.

Explore my work Get in touch Scroll

02 — The idea

Good software doesn't just do more. It removes friction.

A lot of operational work is repetitive rather than difficult. An inquiry arrives by email, the context lives in a spreadsheet, the history sits in a CRM, and someone moves information between them by hand. Thoughtful engineering connects those sources, adds logic where judgment is needed, and lets the workflow run as one coordinated system.

        Stage {{ stageLetter }}
        
        Fig. 02 — From disconnected tasks to a coordinated system

03 — Three ways I build

One process, three disciplines.

AI, automation and software engineering are parts of the same work: understand the problem, design the system, build a practical solution.

A · AI Engineering

Intelligence that works with context.

I build AI systems that retrieve relevant context, use tools, coordinate tasks, and turn unstructured information into useful outcomes.

Retrieval-Augmented Generation
Multi-agent systems
LLM orchestration
Tool calling and integrations
Evaluation and observability

Fig. 03A — Retrieval, reasoning, coordinated agents and validation

B · Automation Engineering

Less repetitive work. More connected systems.

I connect APIs, business tools, data, and intelligent agents to automate workflows that would otherwise require repetitive manual effort.

Business process automation
Email and messaging workflows
CRM and ERP integrations
Event-driven processing
Workflow orchestration

Fig. 03B — Event, processing, decision, existing tools, action

C · Software Engineering

From a clear problem to a working product.

I build complete software solutions, from frontend interfaces and backend services to databases, integrations, and deployment architecture.

React and Next.js
Python and FastAPI
Node.js and REST APIs
MongoDB, PostgreSQL, and Redis
Docker, cloud services, and integrations

Fig. 03C — Interface, API, data and infrastructure as one system

04 — Selected work

Real problems. Thoughtful systems.

Six case studies from professional work and personal projects. Each one explains the problem, the architecture and what changed.

Professional work is described without confidential details. Workflow walkthroughs in each case study are simplified illustrations, not screenshots of deployed products.

05 — How I think

Engineering is a sequence of decisions.

A rough problem statement becomes a working system through a series of choices. I judge those choices by how maintainable, reliable and observable the result is, and by whether the outcome can be measured.

01 Understand the problem. Who does the work today, where it slows down, and what a good outcome looks like.
02 Map the workflow and constraints. Inputs, hand-offs, edge cases, data sources, and the systems that can't change.
03 Design the system architecture. Services, data flow, and where AI helps versus where deterministic logic is safer.
04 Build and integrate. Ship in small increments and connect to the tools people already use.
05 Test, observe, and refine. Tracing, evaluation and measurement, so each change is based on evidence.

    Judged byMaintainabilityReliabilityObservabilityMeasurable outcomes
  

06 — Experience

Where the work happened.

May 2026 — Present

Nickelfox Technologies

SDE 1 — AI Engineer

—Sales Email Bot using RAG and CRM-synced follow-ups.
—Internal ERP/CRM workflows spanning Sales, HR, Finance, Project Management, and Project Vault.
—Contact management and contextual outreach workflows.
—Full-stack client features and AI operations workflows.

June 2025 — April 2026

Boring Workflows

AI Engineer · Internship, remote

—SWARM modular multi-agent architecture.
—Doby legal workflow automation.
—Singtel Chat SDK with a four-stage prepaid sales flow, WebSockets, and BSS Gateway integrations.
—AutoCLM, a configurable AI workflow system using OpenRouter and Claude, with reusable agent templates across five or more projects.