Long Vu Nguyen

Software Engineer

Ho Chi Minh City, Vietnam4 yrs experience

AI EnabledAI ML Practitioner

Key Highlights

Over 3 years of experience in AI-powered systems.
Expert in C and Modern C++ for embedded software.
Proven track record of optimizing system performance.

Stackforce AI infers this person is a Software Engineer specializing in AI and audio processing technologies.

Contact

Skills

Core Skills

C++System OptimizationCross-platform DevelopmentAi-powered ProcessingAi Software DevelopmentAudio Processing

Other Skills

C/C++TclMemory CompilerNodeJSElectronJSPythononnxruntimepulseaudioYocto LinuxRayPytorchllama.cppFastAPIpybind11pffft

About

Software Engineer with 3+ years of experience in developing cloud and embedded AI-powered processing systems. Proficient in cross-platform software development, realtime data processing and system optimization. Specialized in C and Modern C++ multi-threaded programming and embedded software optimization.

Experience

4 yrs

Total Experience

2 yrs

Average Tenure

1 yr 11 mos

Current Experience

Nami technology joint-stock company (namitech jsc)

2 roles

Software Engineer

Dec 2024 – Sep 2025 · 9 mos · Ho Chi Minh City, Vietnam · Remote

Project: NamiAnywhere
Technology: C/C++, NodeJS, ElectronJS.
Participated in the development of NamiAnywhere C++ Client for AI-powered speech processing that work on a diverse number of platforms: on desktop (Windows), on mobie device (Android, IOS, Android-based custom-made device).
Participated in the development of NamiAnywhere Backend.

C/C++NodeJSElectronJSCross-Platform DevelopmentAI-Powered Processing

AI Software Engineer

Jun 2022 – Jul 2024 · 2 yrs 1 mo · Ho Chi Minh City, Vietnam · On-site

Project: CrystalSound - https://crystalsound.ai/
Technology: C/C++, Python, NodeJS, ElectronJS, onnxruntime, pulseaudio, Yocto Linux, Ray, Pytorch, llama.cpp, FastAPI, pybind11, pffft, miniaudio, PyQT5, Ffmpeg.
Software development:
Single-handedly developed a low-cost distributed AI backend in Python and Ray Distributed to translate audio into transcript and insights, the AI backend performs the following AI inference: Audio diarization, VAD, Language detection, and Multi-language audio transcription with Onnx and Pytorch, achieved 0.009 real-time-factor using only one Nvidia T4 GPU, text summarization and analysis via llama.cpp using a self-hosted finetuned large language model based on Llama2-70B.
Participated in the development of CrystalSound C++ SDK for AI-powered low-latency real-time speech enhancement that work on a diverse number of platforms: on desktop (MacOS, Windows, Linux), on browser (via WebAssembly), on several HiFi DSP4 & HiFi DSP 5 audio SOCs.
Participated in the development and maintained audio kernel drivers for MacOS, Windows, and Linux.
Optimized C++ SDKs to reduce latency and CPU usage
Made NamiTech’s real-time target speaker extraction (background voice suppression) demo using Yocto Linux and NXP iMX8 EVK, resulting in good impressions at CES 2022.
Participated in the development of a Window APO kernel plugin for real-time noise cancelling on Windows PCs for an ODM.
Proposed a scheduling method for high performance audio processing on multi-core CPUs.
Maintained, reviewed and refactored desktop ElectronJS app.
Model development:
Developed an UNet-based, dual outputs AI model for acoustic echo cancellation.
Developed post-processing model to reduce residual artifact after noise reduction, improved benchmark speech quality by 3%.
Conducted exploratory research on improving automatic speech recognition with LCMV and MVDR beamforming, improved WER score by 4% compared to the baseline.

C/C++PythonNodeJSElectronJSonnxruntimepulseaudio+12

Synopsys inc

Software Engineering, Senior Engineer

Jul 2024 – Present · 1 yr 11 mos · Ho Chi Minh City, Vietnam · Hybrid

Project: AutoChar (Internal)
Technology: C/C++, Tcl, Memory Compiler.
Effectively provided a throughout code analysis for hidden errors and structural flaws, improved the code quality by 9%.
Optimized C++ library to reduce latency and CPU usage, achieved a single-instance characterization runtime improvement of over 80%.
Conducted exploration code optimization, effectively reduce the IO overhead by 70% and the determination runtime by 10%.
Proposed a caching method for low latency circuit file expansion.
Proposed a merging method for reducing the IO bottleneck during data merge, effectively shortened the runtime by 40%.
Optimized C++ Memory Characterization tools to reduce latency and CPU usage.

C/C++TclMemory CompilerC++System Optimization