Чётко очерченная senior-роль в edge AI даёт возможность работать с реальными устройствами в продуктовой компании и получить долю в компании, но зарплатный диапазон ниже рыночного ориентира, а быстрый темп подойдёт не всем.
Ниже рынка
Высокий темп
Современный стек
Понятные задачи
Сфера ИИ
Мэтч & Сопровод
Для мэтча с этой вакансией нужен Plus
Подключить Plus
Описание вакансии
Текст:
Hirify AI/Оригинал
TL;DR
Applied ML Engineer (Edge AI): Porting Deepgram speech models to non-NVIDIA accelerators, edge servers, and embedded platforms with an accent on quantization, operator substitution, runtime adaptation, and real-device benchmarking. Focus on building repeatable model conversion and deployment pipelines, validating accuracy and latency, and collaborating with embedded engineers and platform vendors on custom kernels and toolchains.
Location: USA — Remote
Base salary: $155,000–$245,000 per year, plus equity, bonus, and benefits.
Company
Deepgram provides real-time speech-to-text, text-to-speech, and voice-agent APIs and deployable voice AI models for developers and organizations.
What you will do
Port speech models to non-NVIDIA accelerators, edge servers, and embedded platforms with minimal model changes.
Make serving-side decisions involving quantization, precision, operator substitution, graph rewrites, and architecture adjustments.
Validate ports on real hardware using accuracy, latency, throughput, and memory benchmarks.
Build repeatable model packaging, conversion, versioning, and automated deployment pipelines for edge targets.
Collaborate with embedded engineers, platform vendors, and silicon partners on kernels, runtimes, and toolchains.
Help address edge deployment concerns such as model security, integrity, automated delivery, and observability.
Requirements
Production experience deploying machine learning models to edge or non-NVIDIA hardware; cloud-only or GPU-only serving experience is insufficient.
Working knowledge of quantization and precision trade-offs, including INT8, FP16, mixed precision, and calibration.
Experience with an edge or vendor inference runtime and conversion toolchain such as ONNX Runtime, TFLite, ExecuTorch, OpenVINO, Qualcomm AI Engine, or a vendor NPU SDK.
Ability to read and rewrite model graphs, replace unsupported operators, and adjust architecture parameters while preserving accuracy.
Strong Python and PyTorch skills, with production-quality testing, reproducibility, benchmarking, and deployment automation.
A builder mindset and ability to scope and deliver ports on unfamiliar platforms.
Nice to have
Experience with speech, audio, or streaming and real-time models.
Exposure to low-level kernels such as CUDA, Metal, NEON, or vendor DSP code.
Experience with model signing, encrypted storage, safe updates, or other deployed-model security measures.
Familiarity with Qualcomm, Apple, ARM, Intel, AMD, or custom NPU accelerator families.
Experience building internal tooling that improves model porting or deployment speed.
Culture & Benefits
AI-first working environment with active use and experimentation with advanced AI tools.
Fast-changing work focused on experimentation, adaptation, and continuous learning.
Opportunity to ship applied machine learning models to customers across multiple hardware platforms.
Equity, annual bonus, and benefits are offered in addition to the base salary.
Будьте осторожны
Если работодатель просит войти в их систему, используя iCloud/Google, прислать код/пароль, запустить код/ПО, не делайте этого. Это мошенники. Обязательно жмите «Пожаловаться» или пишите в поддержку. Подробнее в гайде →