Skip to the archive
Fuad.info
AboutMomentsThermal / Nominal
CV
Index
Engineering archive / vol. 01Baku · Haining · Urbana

Fuad
Safaraliyev

Research / Work / Field notesExplore the archive

ProfilePRF-00OpenSeal

Find the assumption a system depends on, then push it until something interesting happens.

Zhejiang University / UIUC Institute
Dual degree · 2024 – 2028
Junior year at Urbana-Champaign · 2026 – 2027
Open Profile 1 sheet
Sheet 01 / 01

Method + background

Find the assumption a system depends on, then push it until something interesting happens.

That can mean breaking software, questioning a model’s metric, or tuning a control policy until it behaves for the right reasons. I work intensely, but I care about building things, and a life, that can sustain that intensity.

Programme
B.S. Electrical and Computer Engineering, dual degree. Zhejiang University and the University of Illinois Urbana-Champaign Institute.
Standing
3.92 / 4.00 GPA. 60% merit scholarship.
Contact
Zhejiang University / UIUC Institute · 2024–2028End of file / continue through archive

NanoVisionCV-01OpenSeal

Nanovision-EM
Graphene microscopy / image analysis
193
Micrographs
7
Experimental batches
177
QC retained
902
Morphology records
Python pipeline · per-object morphology
Open NanoVision 2 sheets
Sheet 01 / 02

Overview + inspection

Graphene.
From micrograph
to morphology.

193
Micrographs
7
Experimental batches
177
QC retained
902
Morphology records

A Python image-analysis pipeline for graphene micrographs and per-object morphology extraction.

Synthetic inspection sampleNot research data
Synthetic grayscale flake field; not a real microscope image
Synthetic binary mask of the same generated shapes; not a model prediction
50% mask
Reticle / normalized positionX ---- Y ----

The mask comes from the generated shapes, not a model. This pair demonstrates the inspection interface only.

Sheet 02 / 02

Pipeline + notes

  1. 01Contrast enhancement
  2. 02Otsu · adaptive threshold
  3. 03Watershed separation
  4. 04Pseudo-mask generation
  5. 05U-Net segmentation
  6. 06Morphology extraction
Evaluation
0.58 validation IoU against heuristic pseudo masks. This is not final model accuracy or evaluation against hand-annotated ground truth. Hand annotation is still in progress.
Ceiling
Pseudo-mask supervision. The model inherits the heuristics it trained on.
Context
Advisor Prof. Kemal Celebi, ZJUI. University-funded SRTP, May 2025 to present.
Source
Private repository
Python · OpenCV · scikit-image · NumPy · PyTorchEnd of file / continue through archive
RoboArmRL-02Sealed
Roboarm
Reinforcement learning for bio-inspired arm control
Prof. Liangjing Yang · SRTP · 2026 – present
No experimental results yet

Next research direction Adversarial machine learning Under consideration

SecurityADV-04OpenSeal

Structural load / 0.00
Offensive
security
Reverse engineering / cryptography / forensics
Kiberbakmil
Founder and captain
Top-ranked CTF team in Azerbaijan
Two consecutive years
2020 – 2024 · press and drag to load the structure
Open Security 2 sheets
Sheet 01 / 02

Competition + practice

KiberbakmilFounder and captain · Top-ranked CTF team in Azerbaijan · Two consecutive years
2024Cyber Army CTF1st place
2023Codeby platformTop 10 international
2023Cyber Stars3rd national
2020National Informatics Olympiad3rd place
Sheet 02 / 02

Engagement + disclosure

Official mark
not supplied

LeoBank

Public bug bounty · iOS application testing · 2023–2024
Authorised engagements only / no findings publishedEnd of file / continue through archive

Field notes

TXT-05 / Writing
OSINT / CTF write-up

Solving OSINT tasks from CyberColosseum CTF

A walkthrough of five open-source intelligence challenges.

Read on Medium 7 min read

Burn,but do notburn out.

Thermal stateCooling

Contact

Fuad Safaraliyev · Engineering archive vol. 01PrivacyTerms