Ayush JhaAI engineer
RésuméGet in touch

Ayush Jha · AI engineer · Pune

I build AI systems like it’s my own company.

I’m on the AI team at Voiceqube, building multi-agent LLM systems end to end: the agents, the FastAPI services around them and the data underneath. Then I stay for what happens after launch.

  • getPlacedan AI resume tool I built alone, used by 5,000+ people
  • Aligned Automationagents that took over 80% of data validation
Get in touchSee the workRésumé ↗

01Selected work

Three systems, end to end.

Each one at a glance. Run a sample to see how it works, or open the full story.

P-01Live

getPlaced

An AI resume tailor I built alone. Agents rewrite your resume for each job post until it passes the ATS check. Used by 5,000+ people.

Visit getplaced.online ↗Source ↗

getPlaced landing page: Apply less. Land more interviews.
Fig. 1The getPlaced landing page.
How it works

Run a sample request through it, or tap any step. Sample run with a demo resume. Scores are illustrative.

Read the full storyShow less
Problem
Most resumes are rejected by ATS filters in seconds, and tailoring one for every job takes hours.
What I built
A Next.js app on a FastAPI backend that runs a small agent graph: one agent parses the resume, one analyses the job description against a skills knowledge base (RAG), a writer tailors every bullet and a critic scores ATS coverage and sends it back until it passes. Templates export to a recruiter-ready PDF.
Outcome
Designed, built and launched alone. More than 5,000 people have used it.
Built with
Next.js · FastAPI · AI agents · RAG · Firebase

P-02In development

HireLoop

A hiring OS for structured voice interviews. Agents transcribe, follow up and score against the role’s rubric. People make the call.

HireLoop landing page: Structured interviews. Reviewable evidence. Defensible decisions.
Fig. 2The HireLoop landing page.
How it works

Run a sample request through it, or tap any step. Sample run. HireLoop is in active development.

Read the full storyShow less
Problem
Interview decisions rest on memory and gut feel, so they're hard to review and harder to defend.
What I built
A hiring operating system. Candidates take a structured voice interview streamed over WebSockets; a team of agents transcribes, asks follow-ups, scores against the role's rubric (retrieved per role) and pulls out evidence. Every signal lands in one dashboard, and the hiring team makes the call.
Outcome
A freelance build, in active development.
Built with
Next.js · FastAPI · WebSockets · AI agents · Supabase

P-03SIH 2026 · team build

KalaSetu

A voice-first app that gets artisans online: speak, snap a photo and get a fair-priced listing in Hindi and English.

KalaSetu landing page: Empowering Artisans. Enabling Futures.
Fig. 3The KalaSetu landing page.
How it works

Run a sample request through it, or tap any step. Sample run with an illustrative listing.

Read the full storyShow less
Problem
Marginalised artisans and weavers sell at fairs but have no year-round online presence. Many can't type, can't photograph products well and don't know how to price for e-commerce.
What I built
A voice-first app: the artisan takes a photo and describes the piece in their own language. An image studio cleans the photo without changing its colours, a voice cataloger agent asks for one missing detail at a time, and a rules-based pricing assistant suggests a fair price where every number is sourced. The listing is published in Hindi and English, signed and given a QR code.
Outcome
Built with my team for Smart India Hackathon 2026 (problem statement 26090, Ministry of Social Justice and Empowerment): the API, a web catalog and a mobile app in one monorepo.
Built with
FastAPI · Flutter · Next.js · AI agents · Firebase

Want the deeper version of any of these? I’m happy to walk you through the code.

Get in touch

02Revision history

Where I’ve built, newest first.

  1. ENow

    SDE, AI · Voiceqube

    Software engineer on the AI team, building production AI systems.

  2. DJan – Apr 2026

    AI Engineering Intern · Aligned Automation ↗

    Built an agentic validation engine on LangGraph: a supervisor agent routes data to specialist validators and sends failures to a fixer agent, in a loop, until the data is clean.

    • 80% of data validation automated
    • −30% data quality issues via auto-fixes
    • 100 GB+ NASCAR race data preprocessed
    • +40% pipeline processing efficiency
  3. C2026

    Founder & solo engineer · getPlaced ↗

    Designed, built and launched an AI resume builder end to end (Next.js, FastAPI, an LLM agent pipeline, Firebase) and grew it to 5,000+ users.

  4. B2024 – 2025

    Hackathons · 9 competition wins

    36-hour builds: generative-AI agent pipelines, predictive models and data dashboards, shipped under pressure and pitched to judges.

    See all 9 winsHide the list
    • 1stDatabot AI Hackathon · IIIT Kottayam
    • 1stDatathon · SBUP College
    • 2ndMindspark 2025 · COEP Pune
    • WinStartup Ideathon 2025 · JSPM Tathwade
    • WinQuizatron · MMCOE Dexterity 2025
    • WinIIT Goa Datathon · IIT Goa
    • WinCodeVerse 2024 · MMCOE
    • WinDatathon · PICT Pune
    • WinDashboard Competition 2024 · COEP Pune
  5. AAug – Nov 2024

    ML Intern · RHV Group

    Built recommendation models and cleaner sales data pipelines, and presented the findings to clients with Power BI.

    • +15% user engagement from recommendation models
    • +20% analytical accuracy on sales datasets
    • −25% data preparation time via preprocessing pipelines
    • 3 client projects won with data presentations

Education

B.E. Artificial Intelligence & Data Science
MMCOE, Pune · class of 2027 · GPA 9.3 / 10

Competitive programming

LeetCode 1572 · CodeChef 1633 · Codeforces 1308
900+ problems solved

03Parts list

Smaller builds.

From hackathons and weekends. Each one is a part I still reuse: retrieval pipelines, real-time streams, scrapers, prompt-to-UI.

  1. 01

    Cureify

    Multimodal clinical decision support: agents read records, handwritten notes and scans, then propose a next step with a confidence score.

    LangGraph · Multimodal AI · FastAPI

    Live ↗Source ↗

  2. 02

    FinChat

    AI financial analyst: live yFinance metrics, Plotly charts, IPO document Q&A.

    Streamlit · LangChain · LLM

    Live ↗Source ↗

  3. 03

    Vital

    Real-time vitals over WebSockets that predicts patient risk and pages doctors.

    Next.js · WebSockets · Firebase

    Source ↗

  4. 04

    LawBuddy

    Legal RAG agent on the Indian Constitution that answers in plain language.

    LangChain · Pinecone · RAG

    Live ↗Source ↗

  5. 05

    AI Form Builder

    Prompt in, complete survey out: sections, questions, validation.

    Next.js · LLM · Firebase

    Live ↗Source ↗

  6. 06

    Lead Scraper

    Yelp leads across countries, compiled and exported to Excel.

    Python · BeautifulSoup · Pandas

    Live ↗Source ↗

  7. 07

    Tax Minimisation

    Gamified platform teaching students tax, assets and investing.

    Next.js · Node · Framer Motion

    Source ↗

Everything else is on github.com/aj-codespy ↗

Tools I work with

Agents & LLMs

Data & ML

Frontend

04About

The person behind the drawings.

I was born in Bihar, grew up in Mumbai and started building in Pune in 2023, from zero: no web, no cloud, no AI. Two years later I’d won nine competitions, most of them hackathons and datathons, and was writing production agent code.

I like the whole problem, not just the ticket. I sit in on the user call, design the system, write the code and stay for what happens after launch.

Outside of work, I run. And the reason behind all of it is simple: I want to take the money stress off my parents and build them a comfortable life.

The route

  1. 2007

    Bihar → Mumbai

    Moved to Mumbai. Rollercoaster school years began.

  2. 2020

    First clients

    Freelance content writing during lockdown: pricing, deadlines and delivery.

  3. 2023

    Pune, from zero

    Started an AI & Data Science degree with no web, cloud or AI.

  4. 2024

    Under load

    Hackathon wins and a first internship.

  5. 2026

    Shipping

    Production agents at Aligned Automation, getPlaced past 5,000 users, then the AI team at Voiceqube.

For founders & small teams

I also build AI tools for small businesses.

Support bots, chatbots, CRM automations and internal tools, connected to what you already use. Fixed scope, you own the code, and two months of support after launch.

Tell me about your projectOr book a free 30-min call ↗

05Contact

Let’s build something.

Hiring for an AI engineering role, or need something built? Email is the fastest way to reach me.

ajayush2301@gmail.com

Currently SDE, AI at Voiceqube. Open to conversations about AI engineering roles.