Skip to content
0xReso

0xReso

Hi, I am Ayush 👋

data, code, and diet coke.

usually at office, or checking X for latest model releases, or at Rajiv Chowk interchange.

—.————————— years spent

Usually around on XDelhi, India
my little corner of Delhiwork · metro · repeat
A playful animated map of DelhiA city sketch with a moving metro at Rajiv Chowk, India Gate, the Yamuna and trees. A house and bouncing marker in northwest Delhi show Ayush’s current work city. Landmarks are illustrative, not exact office coordinates.Rajiv ChowkIndia GateYamuna ~DELHIsomewhere betweena pipeline & the next train.working around hereDelhi, India ♡N
Company withheld · Data Science & GenAIExplore Delhi ↗

About

Since you are on my website, so let me larp what i do —

frankly, i’ve been working on all sorts of data related work. fetching data through pipelines, automating workflows, creating jobs, then building a GenAI layer over fresh data. creating and running algorithms for A/B experimentation, and sometimes modelling.

i know it felt like linkedinese , but hey, everyone has to make their work sound important, don’t they?

but again, given how much data is being pushed out through AI slop and gems alike, i firmly believe future work will be about managing data at that scale. and i think the datafication is going to be here for a while…

“Data-centric AI is the discipline of systematically engineering the data used to build an AI system.”

— Andrew Ng

Places I’ve worked

  1. Company withheld

    Graduate Engineer Trainee

    Data Science & GenAI department

    Full-time
    Jul 2026 — Present

  2. Company withheld

    Data Science Intern

    Internship
    Jan 2026 — Jun 2026

Projects

built to learn

Deepfake Face Swap

An autoencoder, end to end

Wanted to understand how a face swap actually works. Built the whole pipeline: video frames in, a shared encoder and two decoders, then landmark blending back into video. Checked the results with SSIM, PSNR, and MSE.

Faces
Autoencoder
Blend

shared encoder · two decoders · landmark blending

WorkQueue

Job orchestration for agent workflows

Started with background jobs. Then came jobs that depend on other jobs, and workers competing for the same API quota. Added DAG workflows, a shared Redis rate limiter, and result polling so an agent loop can follow along.

Producer
Redis
Workers

DAG workflows · shared rate limits · result polling

Education

  • Delhi Technological University

    B.Tech, Computer Science

    Delhi Technological University

    8.36 / 10

    2022 — 2026

  • CBSE Class XII

    Sachdeva Public School

    96.80%

    2022

  • CBSE Class X

    Sachdeva Public School

    95.00%

    2020

Say hi

don’t email, just shoot me a message on X. i’m ridiculously active there.

Chat with @0xReso GitHubRead on Medium

What I work with

AI & data

AI engineering

  • MCP
  • Streamlit
  • TensorFlow
  • Keras
  • OpenCV

Data engineering

  • SQL
  • DuckDB
  • Redshift
  • PostgreSQL
  • Redis
  • DBeaver
  • Linux
  • cron

Analysis & modelling

  • scikit-learn
  • LightFM
  • Jupyter

Python toolkit

  • Python
  • pandas
  • NumPy