Ali Naderi

Path

1. 2017Embedded R&DSico Systems
2. 2020Backend systemsDataWare
3. 2022Security & authenticationIDmelon
4. 2024AI engineeringCyberDesign
5. 2025Technical leadershipSciNext

In parallel since 2020: ML research on 3D breast ultrasound, from the MSc at IUST to the Iran Image Processing Lab.

Experience

• 2025 – now
  Chief Technology Officer (AI)

  SciNext · Tehran

  Technical lead of an early-stage AI company. I own the systems end to end: architecture, infrastructure, the AI products on top, and the engineering practice around them.

  • Designed and built the company's production platform on my own, in about a month: self-service GitOps on Kubernetes (k3s) with Argo CD, Gitea Actions and a private container registry. A push to main is built and rolled out with no manual step; new services onboard themselves from a template.
  • Made observability and isolation part of the platform, not per-app work: Prometheus, Grafana, Loki and Tempo for every service; a write-only CI identity for secrets; per-organization namespaces and RBAC; per-application database and object-storage credentials.
  • Turn a business problem into a working system: the products range from a Persian-first chemical compound database with a RAG assistant to data-extraction pipelines and developer tooling, each taken from idea to production.
  • Build with AI as part of the engineering loop. I set the architecture and the constraints, AI agents do the volume, and I review, test and own what ships. One engineer covers ground that used to need a team.

  More

  • The platform runs the company's production services across 5 project organizations. A newly acquired GPU cluster is being brought onto it for training and inference.
  • Technical evaluation and mentorship for early-stage AI startups.

  KubernetesArgo CDGitOpsGitea ActionsPrometheusGrafanaLokiTempoPostgreSQLMinIO

  Read how the infrastructure is built →
• 2026
  Backend Software Engineer

  Symmio · DeFi protocol · Remote symm.io ↗

  • Improved the behaviour of the platform's microservices: reliability and performance under real load.
  • Worked on the anomaly detection stack, built on performance metrics and observability tooling, to catch misbehaving services before users do.

  MicroservicesAnomaly detectionObservability
• 2024 – 2025
  Software Engineer (AI)

  CyberDesign Nordic AB · Stockholm, Sweden

  Indoor positioning from mobile sensor data.

  • Built the machine learning systems that estimate indoor position from mobile sensor data.
  • Built the backend APIs and data pipelines behind model training and deployment.

  PythonPyTorchSensor dataAPIs
• 2022 – 2024
  Software Engineer, then engineering lead

  IDmelon Technologies · Vancouver, Canada idmelon.com ↗

  Passwordless authentication built on the FIDO2 standard. IDmelon later joined HID.

  • Worked on the product from its earliest stage: did the initial R&D, built and demonstrated the prototypes, built the first production version, then kept improving it in production.
  • Designed and built Windows sign-in with FIDO2 security keys for local accounts, which Windows then supported only for business accounts: a C++ credential provider, a background authentication service and a local FIDO server. The first working version took about a week.
  • Led the engineering team as the product matured: designed and implemented major components, helped evolve the architecture, trained and mentored junior engineers, and worked with QC to reproduce, diagnose and fix production bugs.

  More

  • Designed secure authentication systems and distributed cloud services; system-level development, data processing pipelines and API architecture.

  FIDO2Passwordless authenticationWindows credential providerC++Python

  Read the technical case study →
• 2020 – 2022
  Software Engineer (Backend)

  DataWare · Tehran

  • Built scalable backend services and microservice architectures for large-scale data processing.

  MicroservicesBackend
• 2017 – 2019
  R&D Intern → Junior Embedded Engineer

  Sico Systems Intelligizer · Shiraz

  Hardware R&D: FPGAs, microcontrollers and embedded Linux.

  • Built a DTMF detector in hardware, then a GSM/GPRS device around it: it answers incoming calls, plays voice prompts and takes commands and access codes from the caller's keypad. Chose and integrated the chips and made sure the signal quality held up.
  • Evaluated Chinese SoC chipsets and built a Linux-based system-on-board on the Allwinner F1C100S (the Lichee Pi Nano's chip) with analog video input and on-board image processing, such as Laplacian filtering for motion detection.

  FPGAMicrocontrollersGSM/GPRSEmbedded LinuxAllwinner F1C100S

Research projects

• 2026
  Persian text-to-speech with single-shot voice cloning

  Generative speech

  • Transformer-based TTS that speaks fluent Persian and clones a voice from a single sample.
  • Adapted modern neural TTS to Persian: dataset construction, large-scale cleaning, linguistic preprocessing.
  • Built the training and fine-tuning pipelines, plus pre- and post-processing that improve pronunciation and fluency.

  TransformersTTSPersian

  Live demo →
• 2025
  Ez-Crawl: agentic web crawler

  LLM tooling

  • LLM-driven crawler that pulls structured data out of raw HTML with prompt-guided extraction.
  • Runs on local or cloud LLMs, with automated browsing pipelines.

  LLMsAgentsScraping
• 2024
  TDSC-ABUS2023 PyTorch dataset toolkit

  Open source · medical imaging

  • Open-source toolkit for the TDSC 2023 Grand Challenge on 3D automated breast ultrasound.
  • Standard loaders, preprocessing and training interfaces for classification and segmentation of volumetric data.
  • Published on GitHub and PyPI.

  PyTorch3D ultrasoundPyPI

  Read the write-up →
• 2020
  Robotic vision for agricultural automation

  Computer vision

  • Vision system that finds the optimal cutting point on each saffron flower.
  • It became the eyes of a robotic saffron processing machine that cuts the flowers where the model points.

  Computer visionRobotics

  Watch the machine →

Publications

• Current
  Spherical Shape-Based Features for Classification of Breast Tumors in 3D-ABUS Images

  A. Naderiparizi, S. Barekat, E. Kozegar, M. Soryani · Ongoing
• 2025
  Breast Mass Classification in 3D ABUS based on Laplace-Beltrami Spectra and Dual Path CNN

  S. Barekat, A. Naderiparizi, E. Kozegar, M. Soryani · Expert Systems with Applications

  DOI →

Academic service

• 2026
  Peer reviewer

  Journal of Imaging Informatics in Medicine
• 2026
  Judge, national AI research challenge

  Scientific Information Database (SID) and IranDoc

Teaching

• 2022
  Teaching assistant, Computer Vision

  IUST · Dr. Mohsen Soryani

  • Supervised graduate lab sessions.
  • Designed a 3D tumor classification project from the MSc thesis: students built 3D meshes from tumor segmentations and refined them to capture spiculations and narrow protrusions, the surface details that signal malignancy.
• 2017
  Teaching assistant, Data Structures and Algorithm Design

  Shiraz University · Dr. Korush Zyarati

  • Ran quizzes, set and graded homework, designed the midterm and final projects.
• 2016 – 2017
  ACM ICPC representative

  Shiraz University

  • Elected to lead the university's ICPC association: organised contests, taught problem-solving courses, mentored university and high-school competitors.
• 2016
  Teaching assistant, Competitive Programming

  Shiraz University · Dr. Mohammad Taheri

  • Ran problem-solving sessions and coached students for ACM ICPC.
• 2016
  Instructor, Olympiad in Informatics preparation

  Dastgheyb High School (Sampad) · Shiraz

  • Taught high-school students algorithm design and competitive programming for the national and international olympiads.

Education

• 2020 – 2023
  MSc, Artificial Intelligence and Robotics

  Iran University of Science and Technology (IUST) · Tehran

  • Thesis: Classification of human breast cancer lesions in 3D automated breast ultrasound (3D-ABUS) images. Advisors: Dr. Mohsen Soryani, Dr. Ehsan Kozegar.
  • Coursework: machine learning, deep learning, computer vision, image processing, stochastic processes.
• 2015 – 2020
  BSc, Computer Engineering

  Shiraz University

  • Project: designed and built a System-on-Board that runs the Linux kernel and processes analog video for embedded vision.

Skills

Also used: Java, C#, TypeScript, Rust, TensorFlow, YOLO, MySQL.

For language models

This site is written to be easy to read by machines too. /llms.txt is the index: who Ali is, and every page and post with a one-line summary. /llms-full.txt is the whole site, this CV included, as one Markdown file.