Hi, my name is

Aarush Narang

I build  

Data scientist and ML engineer. At HCLTech I shipped a Gen AI copilot that cut mean resolution time 76% (105→25 min) and improved backend throughput 26% across 4,000+ knowledge docs. Currently pursuing an MS in Data Science at the University of Arizona.

View My Work View Resume

About Me

Aarush Narang

I'm a data scientist and ML engineer with two years of industry experience shipping production AI systems. At HCLTech I built Gen AI copilots and Python/C++/TypeScript validation pipelines across 4,000+ enterprise knowledge documents; earlier at Carrier and PwC I shipped ML-driven tooling and graph data pipelines that measurably cut cost and turnaround.

I'm now pursuing an MS in Data Science at the University of Arizona (2025–2027), after a B.Tech in Computer Science from Shiv Nadar University. I care about turning messy data into scalable, measurable decision-making systems.

Programming Languages
PythonC++JavaSQLRGoJavaScriptTypeScript
AI/ML & GenAI
PyTorchTensorFlowOpenCVdlibEasyOCRLLMsAzure AI ServicesAzure OpenAI
Web Development
React.jsNode.jsTailwind CSS
Tools & Platforms
Power BITableauGitServiceNow
Databases & Environments
PostgreSQLMySQLMongoDBNeo4jSnowflakeLinux

Experience

Software Engineer · HCLTech

Jul 2024 – Jul 2025

Generative AI Development Team

  • 26% backend throughput gain and a 76% cut in mean resolution time (105→25 min) by running scenario-based testing on ticket/content-quality time series and shipping Python, C++ & TypeScript validation pipelines for 4,000+ knowledge-base documents.
  • 35% fewer document-quality escalations via what-if analysis on freshness, schema drift, and missing-field scenarios, extending automated validation with anomaly detection.
  • Delivered 100% on-time quarterly safety audits through ServiceNow integrations and reproducible QA exports.
Gen AIPythonC++TypeScriptServiceNow

Software Engineer Intern · HCLTech

Jan 2024 – Jul 2024

Sales Performance Tooling

  • Drove adoption across HR, PMO, and L2 leadership by shipping a React, TypeScript & MS SQL Server application with leadership-ready views.
  • 50% reduction in manual executive reporting effort by engineering reusable SQL features and delivering Power BI & Tableau dashboards.
ReactTypeScriptMS SQLPower BITableau

Web Developer Intern · Carrier

May 2023 – Jul 2023
  • 38% reduction in accounts-payable helpdesk volume (450→280 tickets/week) by integrating an ML-driven FAQ chatbot with validation hooks.
  • 35% increase in in-app engagement by designing React, JavaScript & jQuery workflows with Excel-backed operational reporting.
ML ChatbotReactJavaScriptjQuery

Technical Intern · PwC

May 2022 – Sep 2022
  • Sub-second retrieval for 1,000+ employee records by engineering graph features in Python and validating model assumptions on Neo4j.
  • 40% faster month-end reporting by building Python & REST data pipelines with automated validation.
PythonNeo4jRESTGraph DB

Featured Projects

Computer vision attendance system

Computer Vision Attendance System

Stable camera-to-database latency in pilot runs

Biometric attendance and site-monitoring system: processes field imagery with OpenCV for layout-aware recognition and persists results in PostgreSQL for repeatable comparison testing.

PythonC++OpenCVReactPostgreSQL
Code
Human parsing model output

Self-Correction for Human Parsing

84.7% mIoU on LIP — 2.3% above the published SCHP baseline

Trained and evaluated a PyTorch human-parsing model, engineering input/label pipelines and running what-if scenarios under label noise across specialized evaluation tasks.

PythonPyTorchComputer Vision
Code
Plagiarism detection tool

AI Paraphrase & Plagiarism Detector

Detects machine-paraphrased text from rephrasing tools

Plagiarism tool that flags AI-rephrased passages by combining OCR extraction with the ChatGPT API and a paraphrase-detection pipeline.

PythonChatGPT APIEasyOCRNLP
Code
Publication · DEXA 2024 · Naples, Italy

Analyzing the Efficacy of Large Language Models: A Comparative Study

Peer-reviewed paper (co-authored) presented at the 35th International Conference on Database and Expert Systems Applications, published in the Springer LNCS series.

Read Paper

Let's Connect!