Tempe, Arizona

Sashikant Kumar

Computer Science & Mathematics @ ASU

I build AI products, developer tools, and automation systems, from browser extensions to apps used by real people.

Open to Summer 2027 Software Engineering & AI internships

1st place, ASU Principled Innovation Academy
2 production workflow systems at ASU
Shipped to 15 users across the US and India

View Resume GitHub LinkedIn Email

Projects

Selected software I've built and shipped.

BuiltAI / Information Retrieval

Ambi

Personal Browser Memory & Search

A privacy-conscious browser memory that helps you rediscover pages you've visited, using semantic search, keywords, recency, and browsing context.

Chrome extension that saves useful browsing context locally instead of uploading personal history to a cloud service.
Retrieval combining semantic and keyword search with recency and time-spent signals.
Privacy controls: delete saved data, exclude incognito pages, filter sensitive content before storage.

JavaScriptFastAPIChromaDBChrome Extensions

Source ↗

Technical Details

What I Built

A Chrome extension that saves browsing context locally, and a search layer for finding those pages again.

Architecture

Capture: Chrome extension (JavaScript) that stores browsing context locally, not in a cloud service.
Retrieval: Semantic and keyword search, combined with recency and time-spent signals.
Privacy: Delete saved data, exclude incognito pages, and filter sensitive content before storage.
Backend & data: FastAPI and ChromaDB.

Outcome

Working build. The source is public on GitHub.

Demo / Source

Source: github.com/Sunny-ybe/Ambi-local

ShippedMobile / Full-Stack

ConSnow

Consent-Native Ambient Social App

A location-aware social app that lets your 5 to 10 closest people share everyday context without posting updates. Co-founded, built, and shipped to 15 TestFlight users across the US and India.

15 real TestFlight users across the US and India

FastAPI + MongoDB backend deployed on Railway, using Google Places geocoding to interpret raw location data.
Place learning: Home, Work, Frequent, and Occasional classification.
Location accuracy pipeline: GPS drift filtering, stationary lock, timezone-aware timestamps; iOS (TestFlight) and Android builds.

React NativeExpoFastAPIMongoDBRailway

Technical Details

What I Built

Co-founded and built a location-aware social app for sharing everyday context with your 5 to 10 closest people, with iOS (TestFlight) and Android builds.

Architecture

App: React Native with Expo. iOS (TestFlight) and Android builds.
Backend: FastAPI and MongoDB, deployed on Railway.
Location: Google Places geocoding to interpret raw location data, with GPS drift filtering, stationary lock, and timezone-aware timestamps.
Place learning: Classifies places as Home, Work, Frequent, or Occasional.

Outcome

Shipped to 15 TestFlight users across the US and India.

Built · in daily personal useDeveloper Tools / Productivity

Focus Lock

Daily-Use Productivity Extension

A browser extension I use every day that blocks distracting sites, survives browser restarts, and turns on Apple Focus.

Focus sessions persist across browser restarts.
Locks can't be ended early.
macOS Shortcuts integration enables Apple Focus across devices.

JavaScriptPythonChrome Extensions MV3

Demo video ↗

Technical Details

What I Built

A Chrome extension (Manifest V3) that runs focus sessions and blocks distracting sites while one is active.

Architecture

Extension: Chrome Extensions Manifest V3, written in JavaScript.
Sessions: Persist across browser restarts, and a lock can't be ended early.
Apple Focus: macOS Shortcuts integration turns on Apple Focus across devices.

Outcome

In daily personal use.

Demo / Source

Demo video: loom.com/share/7e273f907d154a8981d91e6de48e0de2

Experience

Systems built for real organizational workflows.

Software Developer, Data & Automation

March 2025 – Present

Ira A. Fulton Schools of Engineering, Arizona State University

I design and build workflow and automation systems for project operations and administrative processes at ASU's Fulton Schools of Engineering.

Project Tracking System

A three-tool project intake and tracking workflow: Microsoft Forms, SharePoint, and Power Automate.
Built for the SSE Data Workgroup.
Live in production.

Purchase & Reimbursement Workflow

Airtable pipeline with 2 public intake forms, 5 relational tables, and 11 status values.
Routing across 3 campus coordinators.
Live per-team budget rollups.
Tested end to end with a real submission, status update, and confirmation email.

ToolsMicrosoft FormsSharePointPower AutomateAirtable

Undergraduate Research Assistant

May 2025 – August 2025

Complex Adaptive Systems Initiative (CASI), Arizona State University

Automated monthly progress audits with Pandas and NumPy and generated visual reports for three student research assistants.
Fine-tuned ResNet18 with FastAI for binary classification experiments and summarized results for a research feasibility assessment.

StackPythonPandasNumPyFastAI

Research & Analysis

Research, experiments, and data-driven investigations.

Multimodal Deep Learning for Lung Cancer Prognosis

Undergraduate research contribution · CASI, Arizona State University · Summer 2025

Question: Which cancer type offers the clearest scientific and clinical case for a multimodal deep learning approach, one that combines what a tumor looks like with what its DNA says?
Method: Literature review of recent multimodal oncology studies, as the only undergraduate on a team of PhD and Master's researchers supervised by Dr. George Poste. The team's proposed framework integrates digital pathology slides of tumor tissue with whole-genome sequencing data from liquid biopsy.
Findings: The team concluded that lung adenocarcinoma (LUAD) was the optimal focus, given the strength of available data and the unmet clinical need. The analysis informed a formal research proposal supporting an institutional partnership between ASU and a publicly traded clinical genomics company, which is in the pipeline.

Read the PDF ↗

More Things I've Built

Experiments and products beyond my main engineering work.

Crash Test Buddy

Prototype

AI Toy Safety Rating Concept

An idea and proof of concept from Hacks for Humanity 2026, the social-good hackathon run by ASU Project Humanities. AI toys talk to children, but toy safety rules only check physical hazards, never what a toy says. The concept is a crash-test rating for them: hundreds of adversarial test conversations grade each toy before it reaches a shop, a safety layer rewrites unsafe answers before a child hears them, and parents scan a QR code on the box to see the live grade. Built with Team Hand in Hand; won 1st place overall and $2,500.

AI SafetyElevenLabsLLM JudgeCloudflare PagesProduct Design

Hackathon proof of concept · Hacks for Humanity 2026

Ember

Prototype

Grief Companion Concept

A grief-companion concept I scoped and led a team to design and prototype. Won 1st place at ASU's Principled Innovation Academy.

Interactive Concept Demo ↗

Portfolio Bot

In Progress

Ask questions about someone's work and get answers grounded in their real projects, with citations.

RAG

Skills & Education

Skills

Languages: Python, JavaScript, TypeScript, C++, Java, SQL, MATLAB
Frontend & Mobile: React, React Native, Expo, Chrome Extensions
Backend & Data: FastAPI, PostgreSQL, MongoDB, Supabase, ChromaDB
Infrastructure & Tools: Docker, GitHub Actions, Railway, Git, Playwright, Pytest, Power Automate, Airtable
Areas: Information Retrieval, RAG, Machine Learning, REST APIs, Automation, Testing

Education & Activities

Arizona State University

Expected May 2028

B.S. Computer Science & B.S. Mathematics · GPA 3.90/4.00

Dean's List every semester · New American University Scholar, Provost Distinction Award

Hackathons: HackAZona, ASA DataFest (solo) · Clubs: SODA, DevLabs, Claude Builder's Club

Let's Connect

I'm interested in software engineering and AI internships, ambitious early-stage teams, and conversations with people building interesting things.

sashikant877@gmail.com LinkedIn GitHub Resume