01AI / Machine Learning
DermoAI
A twelve-class skin-image classifier built on ConvNeXt-Base, evaluated with test-time augmentation. A study in computer vision — and in the limits of a good score.
- Accuracy
- 82.79%
- Cancer sensitivity
- 87.30%
Student · Developer · Maker
IshanAsati.
Exploring the intersection of artificial intelligence, software and engineering. Interested in how things work — driven by what I can build.
A field journal of
projects, experiments
and open questions.
Continue to selected work
Documented, not embellished.
Work in progress, in public.
01 / Selected work
Three projects,
three different edges
of the same curiosity.
Built, broken,
and being built.
02Software Engineering
Atlas Fit
An offline-first strength-training app with deterministic plan generation, session logging, progress tracking and a muscle-focused exercise library. No account, no backend, no network required.
- React, TypeScript and Vite
- Dexie and IndexedDB persistence
- Installable, offline-first PWA
03Hardware / Embedded
Robotics & Electronics
ESP32 experiments with motion, distance, sound and light — including a concept for a calm, physical Study Focus Console. Bench work, honestly labelled.
Read the field notes02 / About
A student,
still collecting
disciplines.
Ishan Asati
India
India
I'm interested in understanding how technology works — from the systems running beneath our software to the models learning from data, and the electronics interacting with the physical world.
I enjoy exploring unfamiliar technologies, building things and learning by experimenting. Most of what I know has arrived through projects that were slightly beyond me at the start.
My interests rarely stay in one place, and that's exactly how I like it. The through-line is not a single stack. It is a preference for understanding the whole system — and for making ideas into something that runs.
- Currently
- Student, building in public
- Based in
- India
- Working across
- AI, software, hardware
- Preferred method
- Build, observe, revise
03 / Currently exploring
The syllabus
is self-set.
A working list, not a skill chart. In use means it appears in something I have built. Exploring means I am still learning my way around it.
- 01
Artificial intelligence
Machine learning, computer vision, neural architectures and local models.
In use - 02
Software engineering
Application architecture, TypeScript, React and the craft of shipping.
In use - 03
Hardware & robotics
ESP32, sensors, displays and small systems that meet the physical world.
Exploring - 04
Systems & open source
Linux, tooling and how software actually meets the machine.
Exploring - 05
Creative programming
Python, Pygame and interactive experiments made for curiosity.
Exploring
04 / Notes
A place for
thinking aloud.
Writing takes longer than building. When there is something worth saying — and worth standing behind — it will appear here.
No notes are published yet. The shelf is intentionally empty.
Planned, not published
- 01Planned
What I am learning about machine learning
Evaluation, failure modes and the distance between a score and a system.
- 02Planned
Building local-first software
Notes from Atlas Fit on state, storage and owning your own data.
- 03Planned
Electronics, from the bench
Sensors, timing, wiring mistakes and the things that only appear in hardware.
Correspondence
ishanlikecube@gmail.com