Profile banner
Can Kocagil

Can Kocagil

AI entrepreneur — Founder & CEO of Tachyo; Finceptor chair

cankocagil
Ankara, Turkey
Joined October 2025
Favicon imagegithub.com
Favicon imagelinktr.ee
Favicon imagecankocagil.medium.com
Favicon imagex.com
Favicon imagetr.linkedin.com
Favicon imagedocs.finceptor.app

Network

5.1K connections

Summary

Computer‑vision & ML researcher–engineer with reproducible public work: Can has multiple open-source repositories and Medium/TowardsDataScience pieces on image captioning, video instance segmentation (TT‑SPN), and neural optimizer implementations. These repos and articles show hands-on research-to-code strengths in vision, sequence modelling, and reproducible experiments. github+2
Web3 founder/operator and product builder: As co‑founder and earlier CEO of Finceptor, Can led product strategy, fundraising and community-facing activities (Finceptor pre-seed announcement and KuCoin AMA), and later transitioned to board leadership while founding Tachyo, an AI product. substack+2
Technical communicator and educator: Can publishes tutorial and research-expository articles (e.g., neural-optimizer from-scratch, image-captioning writeups) to medium/TDS and maintains public code, demonstrating an emphasis on teaching and documenting ML techniques for practitioners. medium+1
Practitioner of applied ML across domains: his work spans academic-style fMRI decoding projects to applied product work (AI wearables prototyping and an AI browser copilot), indicating an ability to move between research prototypes and product-focused engineering. github+1

Work

Education

Projects

Writing

How to vibe prototype AI wearables at home

September 1, 2025

A practical walkthrough of prototyping AI-enabled wearables at home (hardware + software prototyping notes).

Favicon imagelinktr.ee

Neural Network Optimizers from Scratch in Python

November 1, 2021

A tutorial-style article explaining and implementing common optimizers (SGD, Momentum, AdaGrad, RMSprop, Adam) from first principles with NumPy implementations.

Favicon imagemedium.com

Image Captioning by Translational Visual‑to‑Language Models

October 1, 2021

A deep-dive on image captioning approaches, attention, and encoder–decoder architectures applied to vision→language tasks.

Favicon imagemedium.com

Hobbies

Hardware prototyping and AI wearables (occasional hobbyist work and writeups). linktr+1

Open-source research and model/algorithm experimentation (computer vision, fMRI analysis, vision-language). github