Skip to content

$ whoami

+ (01)  Profile(01)

Ayush

Associate Data Scientist

Ayush, @0xReso avatar

Alright, so the about section. Honestly, I'm not obsessed with CS, and I don't want to act obsessed either, so I'll keep it straightforward.

I'm someone who has submitted to the models. Yeah, it's done. Now I just use them to break things apart and connect the dots, irrespective of employment or money. Right now I'm working on prompting, and I want to find out how far I can stress test model outputs, and how fast a model can make me learn things.

The rest of the details about my work, or whatever you call proof of work, are below.

writing: Medium ↗ · Notes ↗

+ (02)  Pipeline numbers(02)
8.3M+
feedback records ingested
2,400+
outlets covered
14
typed agent tools
Weekly
refresh cadence
3
feedback sources merged
6 mo
order history modelled
+ (03)  Index(03)
  • Python
  • SQL
  • DuckDB
  • AWS Bedrock
  • Claude
  • Gemini API
  • MCP
  • Amazon Redshift
  • Streamlit
  • LightFM
ayush@0xreso: ~zsh

ayush@0xreso:~$ fastfetch

ayush@0xreso

role
Associate Data Scientist
dept
Data Science and Innovation
based
Delhi, India
edu
B.Tech (CSE) · 8.36/10
stack
Python · SQL · DuckDB · Bedrock · MCP
shell
zsh · Linux cron
status
Usually around on X

ayush@0xreso:~$

+ (04)  Coordinates(04)
Delhi NCR coordinate map with a metro commuteLine map of Delhi NCR with the Yamuna, an approximate city boundary, and a moving train on a rough metro route linking the markers. Three approximate markers: North West Delhi at 28.71°N 77.10°E, Rajiv Chowk at 28.63°N 77.22°E, and the office near Okhla Bird Sanctuary at 28.55°N 77.31°E.77.00°E77.10°E77.20°E77.30°E77.40°E28.50°N28.60°N28.70°N28.80°NYamunaNorth West Delhi28.71°N 77.10°ERajiv Chowk28.63°N 77.22°Eofficenear Okhla Bird Sanctuary28.55°N 77.31°EN5 km
  • North West Delhi28.71°N 77.10°E
  • Rajiv Chowk28.63°N 77.22°E
  • office · near Okhla Bird Sanctuary28.55°N 77.31°E

Approximate areas, not addresses. Dashed line: rough city boundary. Dotted sky line: rough metro route.

$ cat about.md, About

(02)

I ship production AI that moves real metrics.

  • 01Built an agentic analytics bot: 14 typed DuckDB-backed tools querying 2,400+ outlets via AWS Bedrock, helping drive personalised outlet improvements.
  • 02Weekly-refreshed ingestion of 8.3M+ customer feedback records from three sources, searchable from region down to a single outlet.
  • 03Automated outlet-matching pipeline for smooth experiment runs; these experiments contributed to modest business growth (details on request, DM me).
  • 04LightFM recommenders, outlet clustering, and automated segmentation pipelines.

Beyond that, I’ve spent a lot of time at work scheduling cron jobs and automating workflows. Right now I’m learning how to get the most out of coding agents and building side projects along the way. Currently working on Dadhichi, a prompt library for maximising workflows with these agents.

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

Andrew Ng

$ cat experience.log, Places I’ve worked

(03)
  1. Associate Data Scientist

    Jul 2026 to Present
    Data Science and InnovationFull-timeOnsite
    8.3M+
    feedback records
    2,400+
    outlets covered
    14
    typed agent tools
    1. 01Agentic analytics bot on AWS Bedrock
    2. 02Store.AI / IRA feedback ingestion pipeline
    3. 03Personalised menu sequencing with LightFM
    4. 04Customer segmentation for international markets
    5. 05SQL views, batch ingestion jobs, and cron schedules

    (Python · SQL · DuckDB · AWS Bedrock · Claude · Gemini · LightFM · pandas · Amazon Redshift · Streamlit · Linux cron)

  2. Data Science Intern

    Jan 2026 to Jun 2026
    InternshipOnsite
    1-2%
    intra-cluster variance
    6 mo
    sales history
    Monthly
    segment refresh
    1. 01Outlet clustering with K-Means
    2. 02Experiment cohort pipeline on Amazon Redshift
    3. 03Automated outlet matching for experiments
    4. 04Monthly SQL segmentation pipeline
    5. 05Pricing-reconciliation ingestion

    (Python · SQL · Amazon Redshift · pandas · scikit-learn · psycopg2 · Linux cron)

$ ls ~/projects, Case studies

(04)

Short write-ups of production systems: the problem, the approach, and what came out of it.

  1. (Case study)

    Store.AI / IRA

    Customer feedback ingestion pipeline

    (Python · Gemini · DuckDB · Streamlit)

  2. (Case study)

    Agentic analytics bot

    Plain-English questions, answered with real SQL

    (Python · AWS Bedrock · Claude · DuckDB)

  3. (Case study)

    Personalised menu sequencing

    LightFM recommendations in production

    (Python · LightFM · pandas · SQL)

  4. (Case study)

    Experiment pipeline & outlet matching

    Matched groups, repeatable runs

    (Python · SQL · Amazon Redshift · scikit-learn)

$ ls ~/src, Projects

(05)

$ cat dust.logwins that turned into dust with time

(06)
6742AIR

JEE Mains 2022

All India Rank 6742 in JEE Mains 2022.

10104AIR

JEE Advanced 2022

All India Rank 10104 in JEE Advanced 2022.

2ndRunner-up

Createx Corporate Hackathon

Runner-up at an internal hackathon.

1857Knight

LeetCode

350+ problems solved.

1471Specialist

Codeforces

Specialist rating on Codeforces.

1800+4-star

CodeChef

4-star rating on CodeChef.

$ cat ~/.stack, What I work with

(07)
+ (01)  Agent engineering(01)
  • AWS Bedrock
  • boto3
  • Claude
  • Gemini APIbatched requests
  • MCPModel Context Protocol
  • Agent skills
  • CLAUDE.md · AGENTS.md · Cursor rules
+ (02)  Data & ML(02)
  • Python
  • pandas
  • scikit-learn
  • LightFM
  • SQL
  • DuckDB
  • Amazon Redshift
  • Streamlit
  • Linux cron
+ (03)  Project stacks(03)
  • Java
  • Spring Boot
  • Redis
  • TensorFlow
  • Keras
  • OpenCV

$ cat education.txt, Education

(08)
  • B.Tech (CSE)

    Delhi Technological University, New Delhi

    8.36/10

    2022 to 2026

  • CBSE Class XII

    Sachdeva Public School

    96.80%

    2022

  • CBSE Class X

    Sachdeva Public School

    95.00%

    2020

$ write ~/inbox, Say hi

(09)

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