0%

Aryaman Jaiswal

Undergrad, building for iOS and Android.

I'm Aryaman, an undergrad and a freelancer, building things.

I write native apps for iOS and Android, the AI pipelines that run inside them, and the backends behind them.

SDE InternJune 2026 - Present
Bharti Airtel
Re-architected an internal platform tracking live and historical state across a distributed VM fleet, cutting dashboard query latency by 30%. Built a guardrailed natural-language layer over it: local LLMs, a read-only text-to-SQL agent, and anomaly summaries for provisioning spikes.
Freelance iOS DeveloperJanuary 2026 - Present
SmartoutMedia
Built the SmartOut iOS app from scratch in Swift and SwiftUI, now past 50K installs. It works fully offline across Canada, mapping hunting and fishing regulations onto 456 regulatory zones over a bundled SQLite store.
SDE InternMay 2025 - July 2025
NSUT Exam Cell
Built a platform that automated exam scheduling for 10,000+ students, cutting manual paperwork by 70%. Shipped NSUTrack alongside it, native Android and iOS clients centralizing schedules, attendance, and society notices for 1,500+ students over an offline-first cache.
BachelorsAug 2024 - Present
NSUT, Delhi
surviving Electronics and communications

01WandrTurns a group chat into a decided night out. A seven-stage pipeline pulls intent from a Siri summary, researches live venues, and resolves conflicts down a constraint ladder that relaxes budget before it ever touches dietary or accessibility. The model only ranks; deterministic code alone mints the plan, and the chat itself is never stored.Swift, SwiftUI, Foundation Models, App Intents, Natural Language, Evaluations Framework 02SaldoA student budgeting app that reacts to your balance — a theming engine maps what is left to a palette and repaints the whole interface as the number moves. Receipts are read straight off the camera, goal images are cut from their background on device, and every insight is computed against a store that never leaves the phone. It won Apple's Swift Student Challenge 2026.Swift, SwiftUI, Vision, Natural Language, Foundation Models 03AquaSenseAquarium care where the classifier owns the verdict and the model only explains it — water readings are scored first, so the language layer can phrase a result but never invent one. Three routable specialists split the work: live telemetry, image diagnosis from a photo of the fish, and a remediation cart assembled from whatever just got flagged. It took Best Overall Project and Best Use of Gemini at an MLH hackathon, with a runner-up finish for UI/UX.Swift, Vision, LangChain, Flask, Gemini API, Random Forest, Kotlin

Powerlifting total

Ch1

Deadlift407185 kg

Ch2

Squat308140 kg

Ch3

Bench231105 kg

Master1000

430 kg total

Last 30 days

TracksPlays

ArtistsPlays

119 min listened · updated 2 aug