hi, shobhnik here

23 · he/him · india

open to collaborate
gurugram, IST --:--:--
shobhnik
about

i figure out the system architecture, break down hard problems, and care a lot about code review, basically all the things that still and will always needs a human mind.

most of my code? agents write it. i just make sure it doesn't sucks.

~2 years in. shipped things, broke things, got better.

here are the projects i've built.

contributions
places i worked at
Xane AI

Xane AI

jan 2026 – present

software engineer (backend/ai) · gurugram

  • - Designed and implemented end-to-end system architectures, database schemas, and REST APIs, along with technical documentation and AWS deployment plans for the core product and its microservices serving 3.1M+ users.
  • - Built a multi-tool search agent using DeepAgents (LangChain) with persistent memory via Mem0, backed by hybrid retrieval over Qdrant using BGE-M3 embeddings and Reciprocal Rank Fusion (RRF).
  • - Engineered a document ingestion pipeline handling OCR (MinerU), Markdown-aware chunking, and vector embedding, with image contextualisation for multimodal retrieval.
  • - Explored open-source models, tools, and frameworks, and built internal tools and POCs to validate their fit into existing workflows, including RAGAS-based evaluation of RAG accuracy.
Roger (YC S24)

Roger (YC S24)

may 2026 – june 2026

software engineer · remote · freelance/contract

  • - Built an AI agent that takes a brand's website URL, scrapes its site and online presence, and learns the brand's identity, tone, and visual style to generate on-brand posts, creatives, and UGC-style ads.
  • - Built the publishing loop end-to-end: GPT Image 2 for ad creative generation, Seedance 2 for live-avatar video, and Meta Ads APIs for direct campaign deployment — URL to live ad, fully automated.
  • - Architected the system end-to-end — web crawling, tool orchestration with context and memory management, job queuing and retries, and real-time status tracking for video generation.
Otomashen (E-solutions)

Otomashen (E-solutions)

feb 2025 – jan 2026

software engineer · noida

  • - Built and scaled a high-throughput EDI/X12 processing engine handling 20K+ claim files and millions of encrypted denial records, driving a 166% revenue increase.
  • - Optimized Node.js/TypeScript backend services through query tuning, indexing, and caching, reducing API latency from 10s → <1s.
  • - Engineered a real-time dispatch system for proximity-based doctor matching, request lifecycle (accept/decline), and instant notifications.
  • - Delivered 4 production-grade client systems, leading one end-to-end as Lead Developer, owning architecture, implementation, and deployment.

Stealth Startup (YC-backed)

aug 2024 – jan 2025

software engineer · remote

  • - Owned the product end-to-end as the founding engineer, designing the system architecture, database schema, frontend, backend APIs, and production deployment.
  • - Built a high-throughput AI work management platform using Node.js, TypeScript, BullMQ, Redis, PostgreSQL, LangGraph, and Gemini, processing 1,000+ requests daily across millions of production records.
skills / tools

Languages

JavaScript TypeScript Golang (Go)

Frameworks

Next.js NestJS React.js Express.js Langchain Deep agents

Runtime

Bun Node.js

DevOps / Infra

AWS Azure Vercel Git Docker Github Actions

Databases

PostgreSQL MySQL Redis MongoDB Qdrant Pinecone Clickhouse

AI / Agentic

Claude Code Codex Personalised skill files