On-Device Multilingual Speech AI Engineer

Dev Technology Group, Inc.; Reston (VA); On-site; USD 80,000 - 160,000; Full time

Benefits offered by this job

Generous and flexible time-off policy

Flexible work schedules and telework

Career development opportunities

Industry-leading benefits

401K matching

Team and company social events

Community engagement

Job summary

Dev Technology Group is seeking an applied ML engineer to own the speech-to-text model for a government agency's national security mission. You will adapt large ASR models to run on iPhone, handle code-switching and borrowed terms, and design per-language deployment.

You will move from raw audio data to model adaptation, evaluation, and documentation, partnering with mobile engineering to ensure timing and security. Strong communication to explain tradeoffs is essential.

Qualifications

• Bachelor's degree in CS, DS, ML, or related field.
• Strong data-engineering background building production pipelines for large audio/text data.
• Hands-on experience fine-tuning or adapting speech models with LoRA/QLoRA and compression techniques.
• Practical experience with ASR development and evaluation across languages.
• Strong Python and SQL; experience with PyTorch and Hugging Face tools.
• Experience deploying production ML systems with secure handling of sensitive data.
• Ability to explain model behavior to both technical and non-technical stakeholders.

Responsibilities

• Ingest, clean, label multilingual audio data with attention to code-switching.
• Fine-tune and compress ASR models for on-device memory and latency.
• Design per-language deployment packaging with on-demand weights.
• Handle loanwords and transliteration decisions in transcription.
• Build evaluation pipelines and report results against criteria.
• Document model cards, datasets, and evaluation reports for stakeholders.

Skills

Data engineering

ASR fine-tuning

Python

SQL

Model evaluation

Multilingual

Education

Bachelor's degree in CS/DS/ML

Tools

PyTorch

Hugging Face PEFT

torchaudio

librosa

Job description

Dev Technology Group is seeking an applied ML engineer to own the speech-to-text model for a government agency's national security mission. You will adapt large ASR models to run on iPhone, handle code-switching and borrowed terms, and design per-language deployment.

You will move from raw audio data to model adaptation, evaluation, and documentation, partnering with mobile engineering to ensure timing and security. Strong communication to explain tradeoffs is essential.