Ishan Jarwal — Software Engineer & AI Product Engineer
~/about-me

#Hi, I'm Ishan Jarwal.

// Software Engineer & AI Product Engineer

AI-focused Computer Science student specializing in AI systems engineering, Agentic AI, LLMs, RAG, and scalable software architecture. Proficient in TypeScript, Next.js, Node.js, tRPC, PostgreSQL, Prisma, Docker, and Python, with a strong foundation in DSA, software architecture, and building production-ready AI applications.

Dedicated to engineering resilient, type-safe distributed systems and latency-optimized user workflows. I build software with high performance standards—from sub-millisecond p99 canvas synchronization loops and vector ingestion pipelines to autonomous LLM reasoning agents.

>
“Building software that behaves deterministically under stress is not a luxury; it's a design requirement.”

##Core Skills & Stack

[top ↑]

// Technical capabilities categorized by domain

CategoryTechnologies & Frameworks
*Languages
`C++``TypeScript``JavaScript``SQL``Python`
*Backend
`Node.js``NestJS``Express.js``REST APIs``tRPC``RBAC Auth`
*Frontend
`React.js``Next.js``Redux Toolkit``Tailwind CSS``shadcn/ui``Tanstack Query`
*Databases
`PostgreSQL``MongoDB``Redis``Prisma ORM`
*AI Engineering
`LLMs``RAG``Vector Embeddings``LangChain`
*Tools & DevOps
`Git``GitHub``Docker``Linux``CI/CD``VPS Deployment`
*Software Engineering
`Rate Limiting``Caching``Type Safety``Asynchronous Worker``Message Queues``Error Handling``Input Validation``Modular Architecture``Design Patterns`

##Featured Projects

[top ↑]

// Open-source systems, AI agents, and high-performance applications

Developed a real-time collaborative whiteboard using WebSockets, TypeScript, and HTML5 Canvas, implementing bidirectional communication, multiplayer cursor presence, optimistic UI updates, and interactive drawing primitives.

// Preview: Vizi – Real-Time Collaborative Whiteboard[open demo →]
Vizi collaborative canvas interface preview
Key Technical Contributions:
  • -Developed a real-time collaborative whiteboard using WebSockets, TypeScript, and HTML5 Canvas, implementing bidirectional communication, multiplayer cursor presence, optimistic UI updates, and interactive drawing primitives.
  • -Load-tested HTTP stress-ramp endpoints using Artillery, processing 19,504 requests at 208 ops/sec with 1ms p95 and p99 latency and a 100% successful-request rate in the test environment.
  • -Engineered fault-tolerant client-side drawing workflows with offline mutation queuing, automatic WebSocket reconnection, event deduplication, and Last-Write-Wins conflict resolution to support resilient state synchronization across collaborative sessions.
Stack:`TypeScript``Next.js``WebSockets``HTML5 Canvas``Node.js``Turborepo``pnpm``Artillery``Tailwind CSS``Zod`
role:Lead Developer & Architectyear:2026

Architected a GitHub repository intelligence platform leveraging Retrieval-Augmented Generation (RAG) to enable context-aware codebase Q&A using semantic search and vector embeddings.

// Preview: RepoMemo - Github RAG[open demo →]
RepoMemo GitHub RAG query and AST indexing preview
Key Technical Contributions:
  • -Architected a GitHub repository intelligence platform leveraging Retrieval-Augmented Generation (RAG) to enable context-aware codebase Q&A using semantic search and vector embeddings.
  • -Orchestrated an asynchronous ingestion pipeline to index repository files, generate AI summaries with Batch APIs, persist 768-dimensional vector embeddings for similarity-based retrieval, credit based payment and tRPC for type safety.
  • -Integrated GitHub API via Octokit for type-safe repository access and commit-aware source retrieval, preserving GitHub as the source of truth tied to a consistent commit SHA.
  • -Optimized repository indexing time to ~3ms/file even for larger repositories and query latency to ~5s.
Stack:`TypeScript``LangChain``tRPC``Next.js``TanStack Query``Better-Auth``PostgreSQL``pgvector``Prisma ORM``Polar Payments``Inngest``Zod`
role:AI & Systems Architectyear:2026

Architected an AI Resume Builder SaaS platform with support for 25+ customizable templates and real-time editing, featuring AI content generation, ATS scoring, and vectorized PDF generation.

// Preview: Resumazing - AI Resume Builder SaaS[open demo →]
Resumazing AI Resume Builder interface and ATS match preview
Key Technical Contributions:
  • -Architected an AI Resume Builder SaaS platform with support for 25+ customizable templates and real-time editing.
  • -Implemented AI content generation to assist users in creating resume content and improving resume quality.
  • -Built ATS and Job Fit Analysis features that compare resumes against job descriptions and recommend optimizations.
  • -Built an Autosave feature for data persistence and a vectorized PDF export pipeline for high-quality, print-ready resumes.
  • -Integrated Dodo Payments for payments and subscription lifecycle handling via webhooks.
Stack:`TypeScript``Next.js``React Hook Form``PostgreSQL``Prisma ORM``Google GenAI``Dodo Payments``react-pdf`
role:AI Product Engineeryear:2026

##Education

[top ↑]

// Formal academic qualifications, engineering degrees, and coursework

B.Tech — Computer Science & Engineering

JECRC University
2023 — 2027|distinction:GPA 9.3
Core Engineering Coursework:
`Data Structures & Algorithms``Software Architecture``Database Management Systems``Computer Networks``Operating Systems``System Design`

##Certifications

[top ↑]

// Specialized industry training, system design, and architecture credentials

Software Architecture and Big Data Processing
Instructor / Authority: Michael Pogrebinsky
Oct 2026[Credential]
System Design Masterclass
Instructor / Authority: SweetCodey
June 2026[Credential]

##Get in Touch

[top ↑]

// Direct communication channels, repository links, and social network

I am actively open to discussing high-impact software engineering and AI product opportunities, consulting work, or open-source collaborations. Feel free to reach out directly or send a message below.

>
“Have an interesting project in mind? Let's talk.”
Email Address
// Direct Message Formsmtp: active
* All fields are strictly validated
External Developer Profiles: