01
Research that ships
I have deployed GANs to GCP Vertex AI and built ETL pipelines over real patient records. The gap between a paper implementation and production code is where most of the real work lives — and most of the interesting problems.
Aditya PatelAbout
whoami
Background
I'm Aditya Patel, a machine learning and data engineer doing my M.S. in Computer Science at Stevens Institute, in Hoboken.
I finished my B.Tech at VIT Chennai in 2025 with a GPA of 3.5/4.0, and spent the years around it doing internships that had almost nothing in common — ETL pipelines at Intellect Design Arena, IoT systems at Intuz, and full-stack web at Appuno. That turned out to be the useful part. Each one taught me a different half of what makes a system actually work.
What I care about now is generative models and the distributed data pipelines that feed them. I've deployed multimodal GANs to GCP Vertex AI, built SynMedix on AWS SageMaker to generate 50,000+ synthetic patient records, trained a conditional diffusion model on brain MRI slices, and written Airflow pipelines that regulators' data actually flows through.
The through-line is healthcare AI and the data-scarcity problem underneath it: the most valuable datasets in medicine are the ones you are least allowed to use. Most of my work is some attempt at that. Right now I'm an AI engineer intern at Licent Solutions— that work is confidential, so it isn't written up here — and I'm looking for ML or data engineering roles alongside it. If you're building something in that space, I want to hear about it.
- Based
- Hoboken, NJ
- Studying
- M.S. CS · Stevens · 2025–2027
- Prior
- B.Tech CS&E · VIT Chennai · 3.5/4.0
- Currently
- AI Engineer Intern · Licent Solutions
- Internships
- AI · Data Eng · IoT · Web
- Focus
- Generative models · Data pipelines
- Open to
- Remote · Hybrid · On-site
Timeline2021 — Present
How I got here.
Sep 2021 — May 2025
Education
B.Tech, Computer Science & Engineering
VIT Chennai
Four years of data structures, algorithms and systems programming, graduating with a GPA of 3.5/4.0. The foundation everything else is built on — and the years I learned that understanding why something works matters more than getting it to run.
- GPA
- 3.5 / 4.0
- Focus
- Algorithms · Systems · Applied AI
Jul 2022
Experience
Web Development Intern
Appuno IT Solutions
Built full-stack features for an investor marketplace in ASP.NET MVC and C# — RESTful APIs, role-based access control, responsive UI. My first encounter with code that other people depend on, which is a different discipline entirely from code that merely works.
- Stack
- ASP.NET MVC · C# · SQL Server
- Shipped
- Investor Marketplace Platform
Aug 2023
Experience
IoT Engineering Intern
Intuz Solution
Engineered SHEMS, a smart home energy management system on Arduino and Raspberry Pi: sensor firmware, local data aggregation, and a real-time monitoring dashboard. Hardware teaches you that the data does not simply exist — something physical has to go and measure it.
- Stack
- Arduino · Raspberry Pi · Python
- Shipped
- SHEMS energy monitor
Feb 2025
Experience
Data Engineering Intern
Intellect Design Arena
Designed and shipped the AMFI mutual fund ETL pipeline in Apache Airflow and Python — automated regulatory ingestion, quality gates between stages, and Power BI dashboards for stakeholders. A measurable cut in manual preparation time, and my first taste of infrastructure people quietly rely on.
- Stack
- Apache Airflow · Python · PostgreSQL
- Shipped
- AMFI ETL Pipeline
Sep 2025 — 2027
Education
M.S. Computer Science
Stevens Institute of Technology
Currently at Stevens in Hoboken, focused on machine learning, AI systems and distributed computing. Most of what is on this site was built alongside coursework rather than for it.
- Location
- Hoboken, NJ
- Focus
- ML · AI systems · Distributed computing
Jul 2026 — Present
Experience
AI Engineer Intern
Licent Solutions LLC
Currently working as an AI engineer at Licent Solutions. The work is confidential, so there are no details here.
- Role
- AI Engineer · Internship
- Status
- Ongoing
CapabilitiesBy domain
What I work with.
Machine Learning
- PyTorch
- TensorFlow
- Keras
- Scikit-learn
- GANs
- Diffusion / DDPM
- LLMs
- PaLM-E
Data Engineering
- Apache Airflow
- ETL design
- Pandas
- NumPy
- PostgreSQL
- DynamoDB
- DuckDB
- SQL
Cloud & Infrastructure
- AWS SageMaker
- AWS S3
- GCP Vertex AI
- Docker
- RunPod
- Vercel
- Git / CI-CD
Web & Systems
- Next.js
- TypeScript
- React
- ASP.NET MVC
- C#
- Fastify
- REST APIs
Hardware & IoT
- Arduino
- Raspberry Pi
- Sensor firmware
- Real-time aggregation
Languages
- Python
- C#
- C / C++
- TypeScript
- SQL
How I work
Three things I believe.
02
Read the paper, then question it
I read ML papers on weekends, not out of obligation but because understanding why an attention mechanism works the way it does is the only way to meaningfully adapt it. Copying an implementation is never enough.
03
The boring infrastructure matters most
The model is maybe 10% of the work. Data pipelines, monitoring, deployment reliability — that is what separates a demo from a system people trust. I care about both halves, and the second one is where I have spent more hours.
Off the clock
Cricket, a camera I do not use often enough, and an ongoing and expensive interest in coffee that has produced no measurable improvement in my ability to make it.
- Generative AI
- Distributed Systems
- Healthcare AI
- Cloud Architecture
- Open Source
- Cricket
- Photography
- Specialty Coffee
- Hiking
Next
See the work →