Skip to main content

Available for senior engineering roles

Ankit Patra.
I build work that ships.

Full-stack engineer

Senior full-stack engineer working on AI-powered systems, scalable cloud-native architectures, and motion-driven interfaces with React, Next.js, and TypeScript.

scroll ↓

About

I think like a systems engineer and ship like a product designer.

6+ years across full-stack product engineering — from event-driven AI pipelines and serverless cloud architectures to design-systems, motion-rich interfaces, and Backend-for-Frontend patterns. I plan before I implement, prefer scalable boring tech to clever single-use code, and treat motion as part of the product, not decoration.

2025 – Now

Senior Software Engineer · DocMind · One Convergence Devices

Building a scalable document-intelligence platform on Kubernetes microservices with an event-driven AWS pipeline (S3 → Lambda → Redis → workers).

2025

Senior Software Engineer · Kasu.AI · One Convergence Devices

BFF on Next.js, real-time exception handling, CoPilotKit assistant. Cut AP approval cycle from 1–3 days to under 30 minutes.

2023 – 2025

Senior Software Engineer · DkubeX · One Convergence Devices

Re-architected MLOps platform UI for LLM/embedding deployment. Established Storybook, Jest + RTL, Tanstack Form + Zod foundations.

2022 – 2023

Senior Software Engineer · Rargus · One Convergence Devices

AppSync GraphQL over DynamoDB powering a typed React UI. Cognito auth, Lambda + API Gateway serverless, Fargate jobs.

2020 – 2022

Business Analyst (Frontend Engineering) · Deloitte Consulting India

Browser-based conversational agent (React + Node + Dialogflow) for US-hospital AR workflows. Later: React + .NET features at Deloitte App Studios.

2015 – 2019

B.Tech, Computer Science · VSSUT Burla

CGPA 8.36.

Process

How I build with AI

Cursor and Claude Code are force multipliers — not replacements for engineering judgment. The leverage comes from knowing exactly where to apply them.

01

Write the spec before writing the prompt. If you can't describe the behavior in prose, the model can't implement it correctly.

02

Use AI for the first draft, your judgment for the second. Generated code needs the same architectural review as any other code.

03

Iterate in conversation, not in commits. Architecture decisions are cheaper to refine in a planning session than in a PR.

04

Know when to override. AI optimizes for plausibility. You optimize for correctness in your specific system — trust that.

claude-code
$claude "implement AuthProvider per spec in lib/auth.ts"

Reading spec… ✓

Analyzing existing patterns… ✓

Writing lib/auth.ts… ✓

Done. 3 files modified.

$npm test -- lib/auth_

Planning before prompting. Specs before code.

Principles

Engineering philosophy.

Six things I've found consistently true across every codebase I've worked in.

  • Plan before you code

    Architecture decisions are cheaper in prose than in PRs. A 20-minute planning doc prevents a week of refactoring.

  • Maintainability over cleverness

    The next engineer should understand it without you in the room. Clever is a liability; clear is an asset.

  • Event-driven where it counts

    Decouple producers from consumers. When a system can scale its parts independently, it scales at all.

  • Product thinking

    Every technical decision has a UX consequence. Engineering without product context is just expensive puzzle-solving.

  • Boring is good

    Proven patterns over novel abstractions. The interesting part of the job is the problem, not the framework.

  • Observability first

    If you can't measure it, you can't debug it in production. Tracing and structured logging are not optional.

Stack

Technologies I actually use in production.

No buzzword bingo. Every tool here has shipped something I'm comfortable being asked about in an architecture review.

Frontend

ReactNext.jsTypeScriptFramer MotionShadcn UITailwindRTK Query

Backend & APIs

Node.jsGraphQLBFF (Next.js routes)RESTEvent-driven pipelines

Cloud & Data

AWSKubernetesDockerPostgreSQLRedisDynamoDB

AI / ML Surface

LLM integrationsOCR pipelinesEmbedding workflowsLiteLLMLangfuseDialogflow