jayesh.dev
// Full-Stack Engineer · Agentic AI, Analytics & Cloud

Jayesh Maheshwari

Four years building and scaling a production analytics SaaS — most recently an agentic AI assistant that lets users run the dashboard in plain language. React and Next.js on the frontend, Node.js services on the backend, and PostgreSQL/ClickHouse event pipelines underneath, all running on AWS I owned end to end.

Open to LLM / agentic AI and backend (Node.js) roles — full-time or contract

→ jayeshmaheshwari555@gmail.com→ GitHub→ LinkedIn→ LeetCode
up to 60%
dashboard latency cut
20-30%
query performance gain
555+
problems solved
4 yrs
one product, end to end

About

I'm a full-stack engineer. For four years that meant one product taken end to end: an AI-first analytics SaaS at Fragmatic.io, where I built the Next.js dashboards people actually read, the Node.js services behind them, and the PostgreSQL and ClickHouse pipelines that kept those dashboards fast as event volume climbed.

The part I enjoy is the whole vertical slice — designing the event schema, tuning the query that reads it, shipping the service to AWS, then watching the number land on a chart. Most recently that meant an agentic AI assistant: users describe a segment or campaign in plain language and the agent creates it, calling scoped backend APIs I built for it. Before that, it meant carrying AI-generated recommendations for pages and topics from model output through to something live on a customer's site.

I build with coding agents as part of the normal loop — Claude Code and Codex for test coverage, browser testing and a second pass over my own diffs. They make the feedback cycle shorter; the design calls and the review are still mine.

Away from product work I keep my fundamentals sharp with competitive programming, poke at security for fun on TryHackMe and PentesterLab, and shoot photographs when the light is good.

Journey

  1. Oct 2025 - Sep 2026

    Senior Software Engineer

    Fragmatic.io

    • Built an agentic AI assistant that lets users control the analytics dashboard in natural language — creating segments and campaigns — backed by scoped backend APIs the agent calls.
    • Led migration of high-volume analytics pipelines from PostgreSQL to ClickHouse for large-scale event ingestion and sub-second aggregation.
    • Redesigned the event schema around columnar storage, partition keys, and materialized views.
    • Designed a page-topic clustering system that maps pages to topics and uses AI to generate content and conversion recommendations.
    • Owned the AWS ECS backend for real-time tracking and core services, plus Lambda-based anomaly detection with alerts delivered over SQS.

  2. Oct 2022 - Oct 2025

    Associate Software Developer

    Fragmatic.io

    • Contributed to an AI-first SaaS platform for user segmentation, topic analytics, and dynamic content delivery.
    • Built analytics dashboards with Next.js, Node.js, and Chart.js covering user behavior, traffic trends, and campaign performance.
    • Optimized the PostgreSQL queries behind goal conversions, campaign impressions, and segment profile counts.
    • Implemented engagement tracking for clicks, views, and interactions using Redis and DynamoDB.

  3. Jan 2022 - Mar 2022

    React JS Developer (Intern)

    TatvaSoft

    • Built fully responsive web pages in React from PSD designs, holding the layout across screen sizes.
    • Implemented form validation and data handling on those pages.

  4. 2019 - 2023 | CGPA: 8.01

    B. Tech. in Information Technology

    UIT, RGPV — Bhopal, Madhya Pradesh

    • Coursework: Data Structures, Algorithms, Operating Systems, DBMS, OOP, Computer Networks, Linux Programming.
    • Built a strong base in software engineering and scalable systems.

  5. Completed in 2018

    12th Standard (PCM)

    Mahatma Gandhi Convent Hr. Sec. School — Shajapur, Madhya Pradesh

    • Completed higher secondary education with Physics, Chemistry, and Mathematics.

  6. Completed in 2016

    10th Standard

    Mahatma Gandhi Convent Hr. Sec. School — Shajapur, Madhya Pradesh

    • Completed secondary education with strong focus on science and mathematics.

Impact

  • Built an agentic AI assistant that lets users control the analytics dashboard in natural language — creating segments and campaigns — backed by scoped backend APIs the agent calls.
  • Built comprehensive analytics dashboards using Next.js, Node.js, and Chart.js.
  • Improved key PostgreSQL query performance by 20-30% and optimized event schemas.
  • Reduced heavy dashboard latency by up to 60% via ClickHouse design, partitioning, and materialized views.
  • Developed scalable AWS ECS-based backend for real-time tracking and core service workloads.
  • Implemented anomaly detection with AWS Lambda and alert delivery using AWS SQS.

Stack

$ jayesh --stack32 tools · 6 groups
Languages
  • Python
  • JavaScript
  • Bash
  • SQL
Frontend
  • React
  • Next.js
  • Chart.js
Backend
  • Node.js
  • Dodo Payments
  • REST APIs
  • API Development & Testing
  • Apache Unomi
Data
  • PostgreSQL
  • ClickHouse
  • MongoDB
  • Redis
  • Elasticsearch
  • DynamoDB
Cloud & DevOps
  • Terraform
  • Jenkins
  • Serverless
  • AWS Lambda
  • ECS
  • EC2
  • SQS
  • AWS Batch
  • AWS Glue
AI-assisted dev
  • Claude Code
  • Codex
  • Test generation
  • Browser & E2E testing
  • Code review
jayesh:chill, it's not a real terminal

Work

Agentic AI Assistant · Fragmatic.io

  • An in-product assistant that lets users control the analytics dashboard in natural language — describe a segment or campaign and the agent creates it.
  • Built the scoped backend APIs the agent calls, so each action is a narrow, well-defined operation on the user's own data.
  • Owned it end to end, from the agent's tool calls to the Node.js APIs behind them.
LLM agentsTool callingNode.jsNext.js

PostgreSQL → ClickHouse Migration · Fragmatic.io

  • Led migration of high-volume analytics pipelines from PostgreSQL to ClickHouse for large-scale event ingestion and sub-second aggregations.
  • Redesigned the event schema around columnar storage, partition keys and materialized views.
  • Cut heavy dashboard query latency by up to 60%, on top of the 20-30% gained earlier from PostgreSQL query tuning.
ClickHousePostgreSQLNode.js

Real-time Tracking & Anomaly Alerts · Fragmatic.io

  • AWS backend with an ECS cluster handling real-time user tracking and EC2 instances running the core services.
  • Engagement tracking for clicks, views and interactions on Redis and DynamoDB, powering goal calculations and personalization.
  • Scheduled AWS Lambda jobs detect sudden drops in key metrics and deliver dashboard alerts durably over SQS.
AWS ECSEC2LambdaSQSRedisDynamoDB

PseudoSquad

Visit site →
  • Designed and built the website for PseudoSquad, a five-person product development team, in Next.js.
  • Explored layout and visual directions with Claude Design, then made the calls on structure, copy and interaction myself — AI for breadth, human judgement for what actually shipped.
  • Features an interactive terminal visitors can type into and a week-by-week build plan showing how the squad staffs a project.
Next.jsReactClaude Design

Political Analyser

Repository
  • Analyzed political manifestos to compare ideological patterns across major parties.
  • Applied Python, NLTK, and NLP techniques for sentiment analysis.
  • Presented insights using comparative charts and visual summaries.
PythonNLTKNLPMachine Learning

Beyond

Competitive
Solved 150+ problems on Codeforces · Solved 195+ problems on CodeChef · Solved 210+ problems on LeetCode
Interests
Problem Solving · Ethical Hacking · Gardening · Traveling
Photography
Not a lot, but yes — @soul_flickering
Also on
TryHackMe · PentesterLab