ABOUT THIS ROLEWe are seeking a Senior Software Engineer, Applied AI to own end-to-end delivery of LVT's GenAI-powered security deterrence product. It sits directly on top of LVT's perception stack and turns detections into spoken action in the real world.
Where our AI platform and data teams build the rails, this role builds the product that rides on them and is accountable for production delivery. You will own both the GenAI harness and the application code around it, from design through production and iteration. Because this product acts on the physical world, you'll own the precision and safety bar that comes with it.
This is a hands-on senior individual-contributor role at the intersection of several functions. You'll work cross-team with backend and platform engineers, with the ML/LLM Ops platform you deploy against, and directly with the ML engineers who own the detection models, integrating their work into a shipped product and feeding field behavior back to them. You'll also own pragmatic build-versus-buy decisions for this product: when to self-host versus call a managed model, which voice/TTS approach to adopt, and where to draw the line between product-specific code and shared platform capabilities.
You should be equally comfortable writing production application code, designing an evaluation harness for a system that must rarely act wrongly, integrating against detection models you don't own, and making a defensible build-versus-buy call under real cost and latency constraints.
ROLE RESPONSIBILITIES- End-to-End Product Ownership: Own GenAI-powered security product from design through production and field iteration including architecture, application code, rollout, monitoring, and the follow-through after launch.
- Detection-to-Response Harness: Build and own the logic that turns a detection into the right spoken response including orchestration, prompting, voice generation and customization with a product-level evaluation and regression suite that catches false or mistimed talkdowns before they reach the field.
- Precision & Safety Bar: Own the false-positive and timing discipline a real-world-actuating system demands. Define what "acted correctly" means, instrument it across lighting and scene conditions, and hold the bar as detection types expand.
- Application Engineering: Design and build the services, APIs, and integration code that wrap detection and voice into a product, to LVT's standards for reliability, observability, and operational readiness.
- Cross-Team Integration: Integrate against the ML/LLM Ops serving platform and the data team's datasets and contracts rather than rebuilding them, and partner with the ML scientists who own the detection models turning their models into product behavior and routing field signals back to them.
- Build vs. Buy: Own build-versus-buy recommendations and decisions for this product's components managed model API versus self-hosted, voice/TTS provider versus in-house, third-party framework versus shared platform capability with cost, latency, and maintenance trade-offs made explicit.
OUR IDEAL CANDIDATE- Software Engineering Depth: 6+ years building and shipping production software, backend or full-stack with strong systems and API design judgment and ownership of services in production.
- Applied AI / GenAI Product Experience: Has built products on top of ML/GenAI models including orchestration, prompting, retrieval or tool-calling, and especially generation such as voice/TTS including the evaluation harness needed to keep a non-deterministic system reliable. This is application of models in a product, not model training.
- Real-World / Real-Time Systems: Experience with systems where a wrong output has real consequences real-time, event-driven, or actuating systems and the precision and latency discipline that requires. (security, robotics, IoT, or safety-relevant systems are all relevant)
- End-to-End Delivery: A track record of owning a product or major feature end to end, not just implementing a spec handed down.
- Cross-Functional Collaboration: Effective working directly with ML scientists and across platform, data, and product teams; comfortable integrating against models and contracts owned by others.
- Build-vs-Buy Judgment: Demonstrated pragmatic technology selection under cost and latency constraints, with the reasoning made explicit.
- Technical Foundation: Strong Python, plus the application stack in use (e.g. TypeScript/Node); experience building APIs and production backend systems.
- Education: Bachelor's or Master's in Computer Science, Engineering, or a related field, or equivalent practical experience.
PREFERRED QUALIFICATIONS- Computer-vision / video products, especially detection-driven systems (person/vehicle detection, low-light imaging).
- Voice / TTS generation and customization at production quality.
- Agentic and generative tooling: LangGraph, MCP frameworks, vector databases, and inference/serving platforms.
- Experience building evaluation and regression frameworks for non-deterministic or real-world-actuating systems.
- Familiarity with edge-to-cloud or IoT systems.
COMPENSATIONThe beginning annual salary range for this role is $185,400 - $232,050 USD and is determined by location, job-related experience, and education/training. Your total earning potential is amplified by a bonus structure tied to meeting goals, and you will become an owner from day one through our employee equity program.
BENEFITSWe believe you do your best work when your whole life is supported. We invest in our crew's health, families, and financial futures with a benefits package designed to support you inside and outside the office. Full-time benefits include, but not limited to: Comprehensive health, dental and vision coverage, retirement benefits (401k match up to 4%), and flexible PTO.
