▮ Player 1 - Ready

SHIVAM
SHETH

M.S. Computer Science at Northwestern. I build agentic AI systems and the infrastructure they run on - voice agents, payment rails, NLP screening pipelines, and the systems behind them.

Currently at Xpnse AI + Northwestern FORGE
View ProjectsContact

Select object

Loading_

Based in

Evanston, IL

Focus

Agentic AI + AI Infra

Northwestern

M.S. CS '26 - 3.83 GPA

Off the clock

Guitar / Gym / Football

AGENTIC AI◆AI INFRASTRUCTURE◆LLM SYSTEMS◆DISTRIBUTED SYSTEMS◆FULL-STACK◆DEEP LEARNING◆
AGENTIC AI◆AI INFRASTRUCTURE◆LLM SYSTEMS◆DISTRIBUTED SYSTEMS◆FULL-STACK◆DEEP LEARNING◆

01 / Now

What I'm doing right now

Two roles running concurrently this summer, alongside my master's at Northwestern.

Currently working
XA

Xpnse AI ↗

ACTIVEJul 2026 - Sep 2026

Software Engineering Intern

Evanston, IL - Startup

Owning backend architecture and billing for an expense platform built for volunteer and nonprofit organizations.

  • Architected a Supabase backend with row-level security and multi-org RBAC (admin, treasurer, member), plus SSO/OAuth 2.0 flows hardening authentication, session handling, and data protection across the codebase.
  • Integrated the Stripe API end-to-end for subscription billing - webhooks, idempotency, and failure recovery - owned the CI/CD deployment pipeline, and rebuilt the React frontend for improved UX and accessibility.
SupabaseStripeOAuth 2.0RBACReactCI/CD
NF

Northwestern University — FORGE ↗

ACTIVEJun 2026 - Aug 2026

AI Research Engineer

Network for Collaborative Intelligence - Evanston, IL

Sole technical SPOC for Northwestern Medicine ophthalmology, building an AI applicant screening system.

  • Selected for FORGE, a competitive experiential AI program; serve as sole technical SPOC for the Northwestern Medicine ophthalmology department, driving requirements, deliverables, and stakeholder communication.
  • Building an AI-powered application screening system that automates triage and ranking of ophthalmology-department applicants; designed NLP evaluation pipelines and scoring logic to surface strong candidates and cut manual review time.
NLPApplication ScreeningScoring PipelinesHealthcare AIStakeholder Management
Education
NU

Northwestern University

M.S. Computer Science - GPA 3.83/4.0

Evanston, IL · Expected Dec 2026

Distributed Systems, Data Science, Algorithms for Collective Decision Making, Practicum in Intelligent Systems, AI for Business & Science, HCI.

DJ

D. J. Sanghvi College of Engineering

B.Tech Computer Science and Engineering

Mumbai, India · Dec 2021 - May 2025

Algorithms, Distributed Systems, Operating Systems, Computer Architecture, Software Engineering, Databases, Cloud Computing, Machine Learning.

02 / Quest Log

Completed engineering runs

Internships and AI development work focused on automation and measurable operational outcomes.

Northwestern University

Apr 2026 - Jun 2026

Peer Mentor - CS 110, Intro to Computer Programming - Evanston, IL

  • Mentored 20+ undergraduates through Python fundamentals, debugging, and algorithmic problem-solving.
  • Aligned sessions with instructor learning goals across a wide range of student backgrounds and prior experience.
TeachingPythonMentorship

Dell Technologies

Jun 2024 - Aug 2024

Software Development Engineer Intern - Bangalore, India

  • Fine-tuned and deployed LLaMA and Mistral 7B generative models via REST APIs into CI/CD pipelines, optimising inference throughput and cutting token overhead ~30% across 500+ internal users at sub-second latency.
  • Engineered Python and C++ automation pipelines across 12+ business processes using modular design, Git-based code review, and Agile/Scrum workflows, saving 200+ engineer-hours per month.
  • Integrated JIRA with Power BI via REST APIs for real-time operational dashboards and authored troubleshooting scripts.
LLaMAMistral 7BPythonC++CI/CDPower BI

Jai Bhuvan Developers

May 2023 - May 2024

AI Developer (Part-time) - Mumbai, India

  • Architected a full-stack AI analytics platform in Node.js, Express, and PostgreSQL with RESTful microservice APIs deployed on GCP, boosting demand-forecasting accuracy 15% via time-series ML models.
  • Co-built OpenCV and PyTorch vision pipelines analysing 5,000+ site images, automating defect detection at 87% precision.
  • Built an NLP pipeline processing 200+ contracts, cutting legal review time 40%.
Node.jsPostgreSQLGCPOpenCVPyTorchNLP

03 / Projects

Things I built and shipped

Live products, automation pipelines, and a puzzle game that lets you bet on yourself.

AiDHD - AI Group-Planning Concierge

NEW
AiDHD - AI Group-Planning Concierge screenshot

Next.js - Gemini - ElevenLabs - Webhooks - Supabase - Vercel

A live end-to-end trip and nightlife planner that runs the whole loop: search, cards, pay, confirm.

  • Preferences arrive via WhatsApp and iMessage webhooks; planning runs on Gemini with a live ElevenLabs voice agent painting result cards as tools return.
  • Wired real inventory from Duffel and Ticketmaster with secure Prava checkout - no card numbers ever reach the LLM - behind a Supabase AES-GCM PII vault.
  • Added a reel-to-plan path that OCR-decodes an Instagram reel into a fully costed itinerary.
Live agent ↗

LeanPrompt

LIVE
LeanPrompt screenshot

TypeScript - NLP - Chrome Extension - Next.js

Grammarly for AI prompts. A Chrome extension that rewrites your prompt before you hit send, into a shorter version that means the same thing.

  • Generative AI prompt compression engine cutting token usage 40-75% via semantic pattern matching that preserves intent.
  • Trained on 30,000+ examples across 25 categories, directly reducing LLM inference costs at scale.
  • Surfaces the token, cost, and CO2 saving on every rewrite so the impact is visible in-line.
Live site ↗Try the demo ↗

TrendThread - AI Merch Generator

LIVE
TrendThread - AI Merch Generator screenshot

Python - Flask - LLM APIs - Printify API - Docker

An AI merch generator that takes a trending topic all the way to a published Etsy listing, with a human in the loop.

  • End-to-end pipeline ingesting trending topics, generating designs via LLM and image APIs, and auto-publishing to a live Etsy storefront via Printify.
  • Trend detection to design generation to product publishing, gated by swipe-based human approval.
  • Deployed in a Dockerized cloud environment with CI/CD.
Storefront ↗Website ↗Repo ↗

Amazon Data Scraper: Power BI Pipeline

SHIPPED
Amazon Data Scraper: Power BI Pipeline screenshot

Python - SQL - ETL - Power BI - AWS - pytest

An automated ETL pipeline turning scraped Amazon product data into real-time Power BI dashboards.

  • Ingests 100K+ Amazon product records with pytest-validated transformation logic.
  • Integrated with Power BI for real-time dashboards, reducing manual processing time ~70% end to end.
Repo ↗

04 / High Scores

Hackathon leaderboard

Podium finishes and rankings from ML and AI-focused competitions.

  • 1ST

    DATAHACK 1.0

    Enterprise AI/ML system design and innovation

  • 1ST

    GDSC Winter of Code

    Google Developer Student Clubs build sprint

  • 2ND RU

    National Hackathon

    AI surveillance and patrol optimisation

  • 2ND RU

    Rajasthan Police Hackathon

    AI-driven surveillance and patrol optimisation

  • 3RD

    TECHNOVATE

    Systems-oriented ML prototype built under a tight timeline

  • 3RD

    Data2Knowledge

    Data-to-insight pipeline design and presentation

  • 48TH

    Amazon ML Challenge 2023

    Top 0.5% globally out of 10,000+ participants

Leadership

VP & Co-Founder

DJS ISACA Cybersecurity Chapter

Scaled the chapter to 100+ members and ran 8+ security and systems workshops.

Volunteer Teacher

Teach for India

Taught IoT and robotics to underprivileged students.

05 / Research

Published work

Peer-reviewed and preprint work in applied AI and data systems.

PUBLISHEDFEATUREDIEEE International Conference · 2023

IoT, AI-Enabled Partograph System

Real-time maternal monitoring combining IoT sensors with predictive ML, delivering an 18% improvement in complication detection.

Link coming soon
PUBLISHED2023

Multi-Modal GNNs for Fake News Detection

A graph neural network architecture fusing textual, visual, and temporal-spatial signals, achieving a 12.4% F1 improvement over state of the art.

Link coming soon

06 / Loadout

Equipped tech stack

Languages

PythonGoJavaC++JavaScriptTypeScript

AI + Automation

LLaMAMistral 7BGeminiElevenLabsPrompt EngineeringToken OptimizationSelenium

Cloud + Systems

AWSGCPVercelDockerKubernetesHadoopSparkMicroservices

Databases

PostgreSQLSupabaseMongoDBSQLNoSQL

Backend + Security

Node.jsNext.jsWebhooksStripeOAuth 2.0SSOIAMAES-GCM

DevOps + Testing

GitHub ActionspytestunittestJestPower BIAgile/ScrumJIRA

07 / Player Profile

Outside the terminal

Pick an object to inspect it. The models are generated from pure math, not asset files.

Loading_

08 / Contact

Insert coin to continue

Open to

Software EngineerAgentic AI EngineerAI InfrastructureML EngineerBackend Engineer

Hiring for software or applied AI roles? I'd love to talk.

shivamsheth2027@u.northwestern.edu
LinkedInGitHub+1 (773) 541-1355