System:

Saba Burduli

Software Engineer · AI Engineer

I'm building AI agents, AI tools and open-source software, and publishing every update from my tech career.

Registered to:

Saba Burduli

Tbilisi, Georgia

GitHub · LinkedIn · More about me

Stack:

Languages: C#, C++, C, Zig, GLSL, Python, TypeScript

Architecture: Microservices, DDD, CQRS, Event Sourcing, Event-Driven

Data: PostgreSQL, MSSQL, MongoDB, Cassandra, Redis

Infrastructure: Docker, Kubernetes, RabbitMQ, .NET Aspire, Azure (learning)

Tools & Libraries: Git, Entity Framework Core, Dapper, SignalR, Ocelot, YARP, Hangfire, xUnit, Testcontainers, Playwright, GitHub Actions

AI: Agent runtimes, Tool calling, LLM-powered workflows, Transformers (attention, encoders, decoders), Tokenization (BPE, WordPiece), Hugging Face Transformers, Datasets & Hub, PyTorch fine-tuning

Graphics & streaming: DirectX, OpenGL, RTP, RTMP, HLS, NVIDIA hardware encoding

Experience

• Software Engineer · KinoTechJun 2025 – May 2026
• Software Engineer · CentaureaJan 2025 – Jun 2025
• Co-Founder · CRUD AcademySep 2024 – Present

Full experience

Recent engineering log entries

No log entries yet. New ones will show up here.

See all log entries

Featured projects

• Peek Codex[active]

  A fast, read-only terminal browser for local Codex sessions, built on the supported app-server boundary.

  Go 1.25 · Bubble Tea v2 · Bubbles v2 · Lip Gloss v2 · Codex app-server · source

• Keytomev1.1.0 · [active]

  A fast, offline keyboard-shortcut and command reference for macOS.

  Swift 6 · SwiftPM · macOS 14+ · source

• ACC Setup Coachv0.1.0 · [active]

  A safety-first AI companion for understanding and refining Assetto Corsa Competizione car setups.

  TypeScript · Next.js 16 · React 19 · Tailwind CSS 4 · Cloudflare Workers · live demo

See all projects

What I'm doing now

• learning
  AI / ML Engineering

  • How transformer models work (encoders, decoders, attention)
  • Running pretrained LLMs with Hugging Face Transformers
  • Tokenization (BPE, WordPiece) and training custom tokenizers
  • Loading and preprocessing data with Hugging Face Datasets
  • Fine-tuning pretrained models with PyTorch
  • Sharing models and datasets on the Hugging Face Hub

updated 2026-10-15