About Flash
Flash is an Ohio-based company building AI-powered investigative software purpose-built for law enforcement. The software you work on is used by real investigators and facility staff under real operational conditions.
Location and Type
Remote, US-based. Full-time. Benefits include paid time off.
Position Overview
As an AI Engineer at Flash, you will build and improve the machine learning systems at the core of our platform. You will work in Python on AWS GovCloud, developing the retrieval-augmented generation pipelines, transcription and speech processing, and natural language features that power Flash.
What We Are Looking For
2 years of experience building machine learning or AI systems, with work that has run in production.
Strong Python skills.
- Hands-on experience with large language models and retrieval-augmented generation, including prompt design, retrieval strategy, chunking, and vector search.
- Experience with vector databases and embedding models.
- Familiarity with evaluation methods for model and LLM outputs, so quality is measured rather than guessed.
- Comfort working in a cloud environment. AWS experience preferred, GovCloud a plus.
Eligibility to work in the United States.
Nice to have:
- Speech and ASR experience, including multilingual audio.
- NLP experience such as entity extraction and summarization.
- Prior work in a regulated or security-sensitive domain.
Key Responsibilities
- LLM and Retrieval Systems
- Design, build, and improve retrieval-augmented generation pipelines, including prompt design, retrieval strategy, chunking, and vector search over investigative data.
- Tune and evaluate large language model outputs for accuracy, grounding, and reliability, and reduce hallucination in a domain where correctness matters.
- Work with vector databases and embedding models to keep retrieval fast and relevant at scale.
Speech and Language Pipelines
- Build and improve the transcription and speech-processing pipelines, including multilingual audio and speaker handling.
- Develop NLP features such as entity extraction, summarization, and search across large volumes of call and case data.
Evaluation and Quality
- Define and maintain evaluation sets and quality thresholds for AI outputs, so changes are measured rather than guessed.
- Instrument AI features to catch regressions in accuracy or latency before they reach users.
Integration and Collaboration
- Integrate AI capabilities into Flash's products in partnership with the engineering team, keeping inference cost and latency in check.
- Work with the Site Reliability Engineer to run AI workloads reliably and affordably on GovCloud.
Pay: $85,000.00 - $90,000.00 per year
Benefits:
Work Location: Remote
