Accepted to YC Startup School 2026 · San Francisco, July 25 to 26
Seeking Summer/Fall 2026 internships in AI Product, ML, and Data
Durham, NC · Open to relocate
Jaideep Aher
I turn problems and ideas into shipped products
AI Product Builder | ML Systems | Data
Duke AI master's student building ML systems, data products, and 0 to 1 AI tools. Previously worked on L3 ADAS programs at Magna and in fleet safety, and founded an AI based thrift resale automation company.
Projects
Selected work
Finished products in market across commerce, agri-tech, fleet safety, health, and applied AI, plus Duke coursework and research demos. Built to be used, not only presented.
Course Co-Pilot
Transfer credit evaluation is slow and manual. I built a system that reads transcripts, checks university course catalogs, and flags uncertain cases for another pass so students can get answers in minutes instead of days.
- GPT-4o
- Multi-agent
- Python
- Railway
ThriftSpark
Thrift sellers waste time switching between tools for listings, inventory, shipping, and analytics. I put those workflows into one platform and added AI listing creation so a seller can run the store from one place.
- AI vision
- SaaS
- POS
- Multi-channel
SafeType
Parents often do not see harmful messages until it is too late, but hidden spyware breaks trust. I built an Android safety app that checks messages across major social apps and sends real time alerts while staying visible to the child.
- DistilBERT
- PyTorch
- Android
- NLP
Mira
Chronic illness patients wait months to be seen, and then the visit is short and their history is scattered. I built a pre-visit intake companion that turns messy patient notes into a doctor ready timeline and brief, then audits its own output for missed negation, invented symptoms, and out of order events before showing anything.
- DistilBERT
- Gemini
- NLP
- Next.js
VoiceClone AAC
People with ALS can lose the ability to speak in their own voice. I built an iOS app that clones a voice from a short sample and speaks any typed message back in that person's voice.
- iOS
- Swift
- Voice cloning
- TTS
Agrofix
Many farms spray too many chemicals, which wastes money and harms soil. I built a smart spraying system that targets only the areas that need treatment.
- Computer vision
- Robotics
- Agri-tech
- Embedded
Fleet Kawach
Fleet operators lose money and lives when unsafe driving goes unchecked. I built an AI safety platform that detects risky behavior, helps prevent drunk driving, and coaches drivers inside the vehicle in real time.
- Computer vision
- Edge AI
- IoT
- Fleet safety
Driver Monitoring System
Drowsy and distracted driving cause crashes, but many monitoring tools send raw video to the cloud. I built a privacy first system that tracks fatigue and attention inside the car, sends only compact signals instead of video, and stores alerts for review.
- FastAPI
- MediaPipe
- PostgreSQL
- Docker
- Async
Hundekari
Smallholder farmers lacked a simple mobile tool to manage their business. Shipped a farmer Android app on Google Play with 5K+ downloads that, alongside the web platform, now reaches ~8K unique farmers each month.
- Android
- Google Play
- Agri-tech
HHG Farmer Portal
Farmers often get poor prices because they do not control market data or records. I built a portal with digital receipts, local rate history, and a verified marketplace so they can track prices and sell more directly.
- ERP integration
- Marketplace
- Web
- Agri-tech
BinSnap
Recycling apps often fail on real photos taken near the bin because the images are dark, blurry, or messy. I built a classifier that is more reliable on real world photos, shows confidence, and explains what the model focused on.
- PyTorch
- XGBoost
- MobileNetV3
- ONNX
- FastAPI
- React
Coral Reef Monitor
Coral reefs change faster than divers can inspect them by hand. I built a computer vision system that scores reef health from images so monitoring can scale faster.
- Computer vision
- Deep learning
- Python
- Railway
Duke Reuse 360
Campus dining creates a lot of waste from single use containers. I built a reusable container system that lets students reserve meals, skip the line, and track their impact.
- Full-stack
- AI
- Python
- Railway
Kitchen Verdict
It is hard to show non technical users what model fine tuning actually changes. I built a side by side demo that compares base and fine tuned outputs in real time so the difference is easy to see.
- OpenAI fine-tuning
- LLM
- Python
- Railway
KITTI LiDAR visualizer
Raw LiDAR files are hard to inspect without heavy desktop tools. I built a browser based viewer for KITTI scans so teams can open point clouds, labels, and calibration data quickly.
- Three.js
- WebGL
- LiDAR
- Vercel
MediAlert
Radiology delays can slow down pneumonia diagnosis. I trained deep learning models that flag likely pneumonia cases from chest X rays in seconds.
- CNN
- PyTorch
- Medical imaging
- Deep learning
GitHub
Contributions
Public commit activity over the last year. @jaideep-aher
Rendered by githubchart from public contribution data; it may differ slightly from GitHub's own chart.
Contact
Let's connect
Open to Summer 2026 roles where I can ship AI products end to end, whether in engineering, data science, or product. Reach out by email or LinkedIn. You can also book a time that works for both of us below, or use aher.dev/meet as a standalone link.
Durham, NC, United States