Summary

15 years in software engineering, the last decade on real-time and performance-critical systems in embedded, cloud, desktop, and mobile; much of it low-latency audio.

Co-founder of Roc Streaming (open-source real-time audio stack) and ViveSound (commercial desktop audio engine).

Expertise

• Real-time multimedia: low-latency streaming, synchronization, DSP pipelines, FEC; latency and sync measurement; RTP/RTCP, WebRTC; voice AI integration.
• Audio drivers & APIs: Linux, macOS, Windows; Core Audio, WASAPI, PipeWire, ALSA, AAudio, macOS HAL plug-ins, Windows APO, virtual audio devices.
• Concurrency & performance: lock-free algorithms, real-time safety, profiling, contention analysis; Linux internals.
• Embedded: Linux, Zephyr, cross-compilation; STM32, ARM, MIPS, RISC-V; hardware test rigs, hardware mocks.
• Broader experience: data streaming backends (gRPC, WebSockets); high-load services; native platform integration (Win32, macOS, POSIX; Android JNI/NDK, iOS CF/AVF).
• Languages: C++, C, Rust, Go; working knowledge: Python, C#, Java/Kotlin, Objective-C.

Open Source Work

I maintain around 15 open-source projects of varying scale, spanning real-time audio and engineering tools; some have commercial adopters. The full list is on gavv.net/open-source.

Roc Streaming • Co-founder

2015 - present roc-streaming.org 1.2K+   50+

Open ecosystem for real-time audio over IP, with open-source and commercial users. Co-founded and remain a core maintainer. My contributions:

• Real-time architecture with deadline-aware media pipelines and budgeted task scheduling.
• Adaptive latency, clock synchronization, asynchronous resampling, streaming statistics, lock-free algorithms and allocators subsystem.
• Custom RTP/FECFRAME stack with Reed-Solomon and LDPC-Staircase FEC.
• ADC-to-DAC latency down to 10 ms, sync precision 1-2 ms, and resilient playback with loss recovery over Wi-Fi and Internet.
• OS integrations: macOS virtual device with gRPC control and PulseAudio modules.
• Go and Java bindings (JNI, AAR), native layer for Android (MediaProjection, foreground service), and Kotlin/Flutter bridge.
• Cross-compilation and test infrastructure: QEMU-based CI and hardware acceptance testing across x86, ARM32, ARM64, RISC-V, MIPS.
• Release cycle, code review, documentation, contributor onboarding.

libASPL • Author

2021 - present gavv/libASPL 85+   5+

Open-source C++17 library for building virtual audio devices on macOS.

• Type-safe, extensible abstraction layer over the Core Audio HAL object tree.
• Real-time-safe audio I/O, lock-free control paths, HAL simulation for driver testing.

Signal Estimator • Author

2020 - present gavv/signal-estimator 85+   10+

Audio engineering tool for end-to-end latency and synchronization measurement.

• Measures the complete hardware+software path, from ADC input to DAC output.
• Applies to a wide range of setups: sound cards, network streaming, acoustic paths, virtual devices.
• Computes delay via FFT-based signal correlation, I/O jitter, driver delay, synchronization error.

Founder Experience

ViveSound • Technical Co-founder

2024 - present vivesound.com

Commercial engine for OS-level interception and processing of other apps audio on Windows and macOS. Joined as a freelance engineer, led it to commercial launch, later became an equity partner.

• Owned technical roadmap, release cycle, and QA.
• Cross-platform, multi-process, multi-language architecture spanning background services, virtual devices, and OS integrations.
• Low-latency inter-process audio pipelines with lock-free shared-memory transport and gRPC control.
• Configurable DSP chains with built-in effects, in-process and out-of-process plugins, and on-device and cloud voice AI.
• Native audio (WASAPI, Core Audio) and system layer (Win32, macOS frameworks).
• Customer SDK and documentation (C#, Go).

Freelance Projects

I work independently since 2019. I hold Top Rated status and 100% Job Success Score on Upwork, and have delivered around 10 projects, including a few long-term ones lasting 1-2 years.

Probabilistic NAT traversal for P2P streaming

Hole punching algorithm for a P2P streaming product on embedded Linux.

• Three-axis NAT classifier per RFC 5780, on a custom STUN client.
• Random port probing on the birthday paradox; 1-2 minutes to connect in the hard case.
• Physical network testbed emulating full-cone, port-restricted, and symmetric NATs.
• Integrated into customer’s streaming product, shipped and tested on real ISPs.

Consumer audio device for online music collaboration

Hardware box for jamming online with live instruments; custom PCB, Linux/ALSA.

• Built the audio stack from prototype to production and handed off to in-house engineers.
• Real-time architecture isolating latency-sensitive paths from business logic.
• Embedded Linux peer and cloud multipoint mixer (ARM64, ALSA, DSP).
• P2P streaming with adaptive jitter buffer and forward error correction.
• Tested and measured on a closed-loop calibration rig.
• Achieved end-to-end latency below 30 ms over the Internet.

Wearable audio augmentation for live venues

Wireless device for concertgoers: aligns network playback with the acoustic sound reaching the listener.

• Embedded Linux receiver aligning playback with acoustic watermarks in the captured signal.
• Real-time streaming over Wi-Fi and ALSA I/O with hardware-timestamp delay control.
• Achieved synchronization precision below 0.25 ms.

GPU speech-to-text server with WebRTC

• Real-time self-hosted transcription service based on faster-whisper / CTranslate2.
• Sliding window with segment retirement, making Whisper behave as a streaming model.
• Go/Pion WebRTC with Opus and in-band FEC.

NTP server and client for Zephyr RTOS

• Ported the BusyBox ntpd server to Zephyr RTOS on STM32F7 with hardware RTC.
• Custom SNTP client with IPv6 support.

Wireless audio multi-room software

• Clock-synchronized low-latency streaming for multi-room home audio setups.
• Hardware test rig for measuring sync error between speakers.
• Achieved synchronization precision ~1-2 ms.

Synchronized video monitoring for remote operators

Live video fanned out to operators at distant sites who need to make decisions in lockstep.

• RTSP cameras transcoded to RTP / H.264 with KCP FEC and fanned out to receivers (Go, FFmpeg).
• Custom NTP-like clock sync protocol and fixed-latency frame scheduling (OpenGL).
• Achieved presentation skew below 0.5 ms.

Software hearing aid for Android and iOS

• Real-time hearing augmentation pipeline in a portable C++ core.
• Platform-native low-latency audio with AAudio/Oboe and Core Audio (C++, Objective-C, Java).
• 50 ms mouth-to-ear on budget phones from 2020.
• Latency measured acoustically on a purpose-built hardware rig.

Automated voice call testing over Bluetooth

Lab automation software for scoring voice call quality on mobile phones using PESQ/MOS.

• Python service impersonating HFP/SCO headset for nearby mobile devices (PulseAudio, BlueZ).
• Live streaming via WebRTC for call monitoring, with Opus in-band FEC and PLC.
• Live upload of call recordings to cloud for storage and analysis (GStreamer, FFmpeg, SoX).
• Deployed to lab and got unattended testing running on real phones and carrier networks.

Prior Employment

Team Lead • TradingView

High-load concurrent streaming hub in Go for real-time trading data (30k streams, 20k rps, 400 MBps).

• Led a team of 5-10, designed architecture, performed code review, owned release cycle.
• Rate limiting, backpressure, admission control by memory and contention.
• Debugged performance bottlenecks down to Go runtime: GC pauses, scheduler run-queue latency.

Software Engineer • Zodiac Interactive, InfoTeCS, NTMR

Developed embedded software across digital TV, secure networking, and industrial scanners.

• Embedded Linux for digital TV set-top boxes: video streaming and storage, platform abstraction layer, reproducible cross build (Broadcom, Cisco; ARM, MIPS; uclibc).
• Bare-metal and RTOS firmware: Altera SoC (C, Verilog, UCOS-II), Mikromedia SoC (ARM Cortex-M3), servo and sensor control over CAN, RS-232, UART, live offloading over Ethernet.
• Linux system software: Graphical stack on USB-connected microcontroller, serial-over-IP for remote industrial equipment, distributed computing on MPI cluster.

Technical Writing

Full list on gavv.net/articles.

• Roc Toolkit docs — 50 pages on pipelines, threading, timestamp domains, FEC, resampling.
• PulseAudio under the hood — 150-page book; referred by the official PulseAudio documentation.
• Working on a new network transport for PulseAudio and ALSA — Design and benchmarks of the first release of Roc Toolkit.
• File locking in Linux — BSD, POSIX, and OFD locks: semantics, trade-offs, examples.