Mahansh AdityaSoftware Engineer
Bengaluru, INDHeadout · since Aug 2025Available for freelance work

One engineer, the whole stack.

Freelance builds — web apps, storefronts, internal tools — taken from architecture to interface, by one person who does everything.

Now
Software Engineer at Headout
API optimization, service architecture, observability
Stack
Spring Boot · Golang · Next.js · Node.js
PostgreSQL, Kafka, Redis, AWS
Shipped
Anarchist (opens in a new tab)
D2C storefront and CMS, built solo
01

The short version

Two years spent where the load is: queues, jobs, and the dashboards that depend on them.

B.Tech CSE with a Specialisation in Design, IIIT Delhi. The design half shows up in the tools other teams use every day — admin panels, flow editors, CRMs.

Works on
Async job orchestration, event ingestion pipelines, data freshness, notification delivery
Reaches for
Spring Boot, Golang, PostgreSQL, Kafka, Redis, Next.js, Node.js
Also happy doing
Full-stack product work end to end — architecture, API, and the interface on top
  • 70%

    API latency cut by relocating backend and database closer to users

  • 1 day → hrs

    Chapter setup time, after a visual branching-flow editor replaced JSON config

  • 25%

    Faster initial load from lazy-loading and deferred API calls

02

Where the work happened

03

Selected work

Aug 2026/Deployed

Anarchist

A full-stack D2C e-commerce platform with its own CMS panel, architecture and build owned solo — then tuned for multi-region deployment.

  • Next.js
  • Spring Boot
  • Kotlin
  • PostgreSQL
Region latency · try it

Before · US-East origin

baseline latencyevery call crosses an ocean

Backend and PostgreSQL relocated closer to Indian users, with the storefront served from the Vercel edge network.

Aug 2025/Deployed

Cue

A social media-tracking platform for movies and TV shows — with personal and shared watch tracking, rewatch history, progress tracking, ratings and favorites.

  • Next.js
  • PostgreSQL
  • TMDB API

Four things Cue does · try them

Season 1 · episode grid
Season 1 · in progress5 / 16 episodes

Progress is stored per user and per season, so a shared watch keeps both people’s positions separate.

Nov 2024/Deployed/GitHub

mahansh.AI

An AI conversational platform with RAG-based long-term memory — OpenAI embeddings and semantic search over pgvector for context-aware replies.

  • Next.js
  • pgvector
  • OpenAI embeddings
Context sent to the LLM
Token-aware retrieval14 / 48 chunks sent

Semantic search over pgvector picks only the chunks that matter, cutting tokens and API cost while keeping relevant context.

04

Stack

Hover a group to read what it is used for. Tap on a touch screen.

Foundations4 skills
  • DSA
  • OOP
  • REST APIs
  • Distributed sys.
Frontend3 skills
  • React.js
  • Next.js
  • JavaScript
Backend9 skills
  • Kotlin
  • Spring Boot
  • Golang
  • Java
  • Python
  • C / C++
  • Node.js
  • Express.js
  • FastAPI
Database4 skills
  • PostgreSQL
  • MongoDB
  • Redis
  • BigQuery
Cloud & tools4 skills
  • AWS
  • Docker
  • Git
  • GitHub