Skip to content
View eminsk's full-sized avatar
3 followers Β· 0 following

Achievements

Achievement: Pull SharkAchievement: QuickdrawAchievement: YOLO

Achievements

Achievement: Pull SharkAchievement: QuickdrawAchievement: YOLO

Block or report eminsk

Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
eminsk/README.md

Hi there, I'm M_N_Nik πŸ‘‹

Software Engineer & Systems Developer with an engineering background spanning back to the early 2000s.

Combining decades of foundational systems programming β€” from low-level architectures (x86/x64 Assembly, Win32 API, C) to modern high-performance Python engineering, asynchronous services, desktop applications, and algorithmic data pipelines.

Telegram Badge Email Badge

🧭 Engineering Philosophy & Background

  • Decades of Craft: Writing code since the early 2000s, built on a deep understanding of memory management, operating system internals, and native Win32/POSIX runtime behavior.
  • Hardware-Aware Performance: Bridging high-level Python productivity with bare-metal speed β€” leveraging custom SIMD SSE2 vectorized math kernels, FASM assembly, and zero-copy C extensions.
  • Architectural Discipline: Strong emphasis on clean code separation, strict typing (mypy), modern tooling (uv, ruff), reproducible builds, and native assembly compilation (FASM x64).

πŸ›  Tech Stack & Core Competencies

  • Languages: Python (3.10–3.15), x86/x64 Assembly (FASM), C / Win32 API
  • Concurrency & Internals: Asyncio, Threading, SIMD Vectorization (SSE2/AVX), ctypes & C FFI, Inter-Process Communication (IPC)
  • Desktop & UI Engineering: CustomTkinter, Flet, Tkinter, ttkbootstrap, High-DPI scaling, Native Win32/FASM x64 engineering
  • Data, Math & Trading: OpenPyXL, Pandas, TA-Lib (technical analysis & pattern recognition), NumPy, SQLite
  • Web, Scraping & Automation: Playwright, Requests, cURL-cffi, Pydantic, REST APIs, Session & Auth management
  • Quality & DevTools: uv, ruff, mypy, pytest, Git, GitHub Actions CI/CD

πŸš€ Featured Projects

  • agentjit β€” Just-In-Time Compiler for AI Agent Trajectories. Compiles flaky, 30-second multi-step LLM workflows into 0.08ms deterministic Python code with zero token cost. Features speculative guards, automatic bailout runtime, and free-threaded (No-GIL PEP 703) support.
  • nanorecall β€” 100% Private, Zero-Cloud, Bare-Metal Desktop Memory & Screen Search Engine in <200KB. Runs on standard CPUs with zero telemetry, automatic privacy shielding for password managers, and sub-millisecond semantic search via NanoVector.
  • nanovector β€” The SQLite of Vector Search & Episodic Memory for AI Agents in ~120KB. Zero dependencies, instant 0.2ms cold start, pure C99 with AVX2+FMA, ARM NEON, and hand-crafted FASM x64 assembly kernels.
  • nanogemm β€” Minimalist, bare-metal AVX2+FMA & ARM NEON matrix multiplication engine in ~100KB for Python, outperforming NumPy by up to 2.8x on CPU for edge neural network inference. Featured in The Daily Diff (9/10 Depth Score), trending on GitHub (Deep Learning), and published on PyPI.
  • screenvideo β€” High-performance desktop screen recorder and screenshot suite featuring WASAPI loopback audio, real-time H.264 streaming, and a standalone pure x64 Flat Assembler (FASM) native edition (Release v2.0.0).
  • xlsx_vievers β€” High-performance spreadsheet library and desktop processor with a headless 129-function formula engine, hardware-accelerated SIMD SSE2 math, Chart Wizard, AutoFilter, and Goal Seek. Published on PyPI as xlsx-viewer-pro and packaged on conda-forge.
  • yfinance-ta-patterns β€” Algorithmic candlestick pattern scanner, backtesting and AI ranking suite powered by TA-Lib and yfinance. Published on PyPI with Python 3.12–3.15 support and precompiled native wheels.
  • ppa β€” Official APT Repository (PPA) for Debian/Ubuntu packaging agentjit, nanogemm, nanorecall, nanovector, xlsx-viewer-pro, and yfinance-ta-patterns.
  • awesome-baremetal-ai β€” Curated benchmark leaderboard and directory of ultra-lightweight, zero-dependency C/C++/Rust/Assembly engines for local LLMs, edge inference, vector search, and AI agents.
  • StackOverflowAPI β€” Modern bilingual desktop client for Stack Overflow built with CustomTkinter, Catppuccin palettes, syntax highlighting, offline bookmarking, and native FASM x64 search client.

🌐 Open Source Contributions

πŸ“¬ Get in Touch

Pinned Loading

  1. xlsx_vievers xlsx_vievers Public

    Desktop spreadsheet processor: 80+ formula functions, Chart Wizard, AutoFilter, SIMD SSE2 math engine and Win32 GUI

    Python 1