Open to AI Engineer & Full-Stack roles

Hi, I'mAyush Dixit.AI Engineer & Full-Stack Engineer

$AI Engineer

I make language models useful — grounded in real sources, wired into real systems.

I build AI products end to end — the retrieval, the memory, the streaming UI, and the infrastructure underneath. Five years of enterprise engineering taught me how to make systems that hold up: hybrid retrieval over pgvector and Postgres FTS, MCP tool servers that let models act, layered memory so context survives the conversation, all wrapped in Next.js and TypeScript and shipped on Docker and Kubernetes I run myself.

  • RAG & Retrievalpgvector · FTS · RRF
  • Agents & MCPtool servers · memory
  • Full-Stack DeliveryNext.js · Spring · K8s
Noida, India
Ayush Dixit — AI Engineer and Full Stack Engineer
AI ENGINEER · FULL STACK · RAG & LLM SYSTEMS · AD
IngestEmbedRetrieveReasonShip
0+

Years Experience

0

Projects Shipped

0

Public Repos

0

Companies

RAGpgvectorOpenAI APIMCPEmbeddingsGemini APITypeScriptNext.jsReactNode.jsJavaSpring BootPostgreSQLMongoDBRedisDockerKubernetesAWSGraphQLVector SearchMicroservicesRAGpgvectorOpenAI APIMCPEmbeddingsGemini APITypeScriptNext.jsReactNode.jsJavaSpring BootPostgreSQLMongoDBRedisDockerKubernetesAWSGraphQLVector SearchMicroservices

01How I Think

Engineering principles

A working manifesto — tap through the six ideas that shape every system I ship.

01
01 / 06

Architecture first

I design for the boundary before the line of code — clear layers, explicit contracts, and data flow you can reason about. Good structure makes everything after it cheaper.

02Career

Work experience

~5 years across telecom, banking, and enterprise commerce — designing and shipping production systems as a software engineer and full-stack developer.

HCL

Senior Software Engineer - I

Current

HCL Software

Apr 2025Present
1 yr 5 mos Pune
  • Developing cloud-native microservices for HCL's commerce platform using Java, Spring Boot, and Redis
  • Designing scalable REST APIs with Spring MVC and Spring Data JPA
  • Containerizing microservices with Docker and deploying on Kubernetes
  • Building responsive React frontends and SEO-friendly server-rendered pages in Next.js
  • Implementing secure authentication with JWT and Spring Security
  • Applying Redis caching for faster data retrieval and session optimization
  • Conducting code reviews, unit/integration testing, and following Agile/Scrum
JavaSpring BootRedisDockerKubernetesReactNext.js
AC

Software Development Senior Analyst

Accenture

Oct 2024Apr 2025
6 mos Noida
  • Built a secure digital onboarding module with encryption flows for the banking domain
  • Developed scalable Spring Boot microservices and secure REST APIs (Spring Security + JWT)
  • Implemented Spring Batch jobs for large-scale data processing
  • Used Quartz Scheduler for recurring automated reporting jobs
  • Followed Agile practices: sprint planning, standups, retrospectives
  • Conducted code reviews and authored JUnit/Mockito unit tests
JavaSpring BootSpring SecuritySpring BatchQuartz
AM

Software Developer

Amdocs

Aug 2021Oct 2024
3 yrs 2 mos Gurugram
  • Contributed to a telecom Ordering Management System (OMS)
  • Implemented business logic with Java, EJB, and WebLogic Server
  • Integrated SOAP web services and internal APIs; handled async messaging with JMS
  • Wrote and tuned Oracle PL/SQL for large database operations
  • Participated in Agile SDLC and collaborated with QA, DevOps, and Business Analysts
  • Used Jenkins, Maven, Git, and JIRA for CI/CD and version control
  • Performed debugging, unit testing, and production issue resolution
JavaEJBWebLogicSOAPJMSOracle PL/SQLJenkins

03Selected Work

Projects with depth

Not just screenshots — the problem each one solves, how I reasoned about it, the system design, and the real tech stack pulled from GitHub.

01Live2026· Creator

ChaiBook LM

A source-grounded research workspace with citations and long-term memory

A NotebookLM-style research workspace: ingest PDFs, websites, YouTube, email, and MCP snapshots into isolated notebooks, then ask questions that stream cited answers. Hybrid retrieval (pgvector + Postgres FTS + RRF), dual-voice podcasts, YouTube roadmaps, and layered memory so the model does not lose the thread.

Hybrid retrieval you can inspect — RRF, vector, and FTS scores per cited chunk
Layered memory: Redis STM, pinned facts, episodic embeddings, graph, optional Mem0
Sources from PDF, URL, YouTube playlists, email, calendar, Drive, and MCP snapshots
Production deploy: Traefik gateway, private Caddy, unpublished Postgres and Redis
02Live2026· Creator

Shirube

The TypeScript SDK for production AI agents

A TypeScript agent SDK (npm: shirube-ai) that owns the loop after you define an agent, tools, and a model — guardrails on by default, Zod tools with fail-closed approval, handoffs that cannot loop, layered memory, a living knowledge graph, MCP both ways, and streaming events. Security is on unless you opt out.

Security on by default: jailbreak, injection, PII, secret leak, system-prompt leak
Five memory layers kept apart — session ≠ long-term notes ≠ typed graph
Handoffs with a visited-set and maxHandoffs so A→B→A cannot loop forever
MCP both ways, model fallbacks (OpenAI → Claude → Gemini), structured Zod output
03Open Source2026· Creator

Clipo

A fast, native, privacy-first clipboard manager

A lightweight clipboard history manager for macOS and Windows built with Tauri v2 + Rust + SQLite (FTS5) and a React + TypeScript + shadcn/ui frontend. Search, pin, tag, edit and paste back anything you've copied — 100% local.

Native binary, ~10–15 MB installer, cold start under a second
Full-text search (FTS5) that stays fast past 100k+ entries
Pins, favorites, tags, notes, and edit-in-place
Global hotkey + system-tray; 100% local, no telemetry
04In Progress2026· Full-Stack Engineer

Nakshatra AI

AI-powered Vedic astrology platform

A monorepo combining a Next.js 16 web app with a Python FastAPI astro-engine microservice — kundli generation, kundli matching, an AI astrologer chat, and daily horoscopes, with auth, payments, and an admin dashboard.

Polyglot monorepo: Next.js 16 + Python FastAPI microservice
NextAuth v5 with Email OTP + Google and role-based middleware
Dual-LLM AI chat (Gemini primary, OpenAI fallback)
Razorpay payments and an admin dashboard
05Live2025· Full-Stack Engineer

mAI-school

Multi-tenant school management SaaS

A containerized, multi-tenant school-management platform. Every institute signs in on its own subdomain with its own branding; admins, principals, teachers and students each get a focused, role-based home — attendance, fees, exams, announcements, meetings, and AI assists.

True multi-tenancy: subdomain + branding per institute
PostGraphile auto-generates the GraphQL API from the DB schema
Role-based access for admin, principal, teacher, student
Self-serve onboarding with emailed sign-in links; Dockerized
06Open Source2026· Creator

PostGraphile Java

Auto-generate a GraphQL API from your PostgreSQL schema

A Spring Boot starter that introspects your PostgreSQL database and generates a full, Relay-compliant GraphQL API — the Java equivalent of PostGraphile. Zero boilerplate: tables, views, FKs, enums and functions become types, queries and mutations.

Zero-config: PostgreSQL schema → Relay GraphQL API
DataLoader batching eliminates N+1; keyset cursor pagination
Per-request JDBC transaction + PostgreSQL row-level security
Smart Comments, full-text search, JWT auth, virtual threads
07Open Source2026· Creator

MCP Mermaid ER Server

Turn an ER diagram into a live REST + GraphQL API

An open-source Model Context Protocol (MCP) server that parses Mermaid ER diagrams, creates PostgreSQL tables, and exposes automatic REST and GraphQL CRUD APIs — driven entirely by an AI agent through 12 MCP tools.

Model Context Protocol server with 12 agent-callable tools
Mermaid ER → PostgreSQL DDL with PK/FK/UK detection
Auto-generated REST and GraphQL CRUD endpoints
Drives end-to-end from diagram to running API

05Toolbox

Skills & tech stack

A living map of the tools I reach for — filter a layer, or scan the whole stack at once.

$ ls ~/toolbox

56 tools · 8 layers

AI Engineering

10
RAGVector SearchEmbeddingspgvectorMCPAgent ToolingLLM StreamingPrompt EngineeringKnowledge GraphsMem0

Model APIs

05
OpenAI APIGemini APIFunction CallingStructured OutputsZod

Languages

06
TypeScriptJavaScriptJavaSQLHTMLCSS

Backend & APIs

08
Node.jsExpressSpring BootSpring SecuritySpring Data JPAMicroservicesGraphQLREST

Frontend

07
ReactNext.jsTailwind CSSshadcn/uiFramer MotionZustandTauri

Databases

05
PostgreSQLMongoDBRedisOracleSQLite

DevOps & Cloud

08
DockerKubernetesAWSJenkinsCI/CDCaddyTraefikGit

Quality & Tooling

07
ZodJUnitMockitoJIRAAgileMavenGradle

06Background

Education & certifications

My academic foundation and the certifications I've earned along the way.

MMIT, Pune

Bachelor of Engineering in Computer Science

July 2017June 2021CGPA 8.0

Certifications

01Java Spring Framework with Spring Boot – Udemy
02Java Certification – HackerRank
03Python Programming – Microsoft
04Data Analytics – IBM
05Microservices with Spring Boot – Coursera
06Docker & Kubernetes – Udemy