PHYSICAL AI / AUTONOMY / REAL-WORLD OPERATIONS
ON-PREMISE / EDGE AI LEAD
Location : Austin, USA / Hybrid
ABOUT CENTINUS
Centinus builds physical AI and autonomous systems for real-world operations where speed, reliability, and operational effectiveness matter.
Our technology brings AI into the physical world, helping organizations understand complex activity, make faster decisions, and operate intelligent systems in demanding environments. We build for performance beyond the lab - where systems must be dependable, scalable, and ready to support meaningful action in the real world.
We are assembling a small, senior engineering team to solve difficult problems with lasting impact. This is an opportunity to shape the product, the engineering foundation, and the culture of an ambitious early-stage company.
THE ROLE
Centinus is seeking an On-Premise / Edge AI Lead to take our perception, world-model, and reasoning stack out of the cloud and run it on hardware the customer controls: a single GPU server, a ruggedized workstation, or an embedded module on a vehicle or drone, with no external connectivity.
You will decide what runs where, redesign services and models to fit strict memory, compute, power, and latency budgets, and shrink large models, including LLMs and vision-language models, without losing the accuracy that matters operationally. Your work determines whether a capability that runs on a cloud GPU today can run at the tactical edge tomorrow.
This is a hands-on lead role with broad responsibility across edge architecture, model optimization, and on-hardware validation.
WHAT YOU WILL DO
Define the edge architecture: how detection, tracking, re-identification, enrichment, and reasoning split across standalone, on-premise, and embedded targets, and what each target must deliver.
Redesign services and runtimes for constrained, disconnected systems, with predictable memory use, scheduling, and failure behavior, graceful degradation, and no dependence on external services.
Optimize and compress models for the target hardware using quantization, pruning, architecture changes, and compiler and runtime tuning, to meet explicit memory, compute, power, and latency budgets.
Distill large LLMs, vision-language models, and foundation models into compact models that preserve task accuracy, including the teacher-student pipelines, data, and evaluation needed to prove it.
Build the benchmarking and hardware-in-the-loop validation that sets budgets, catches regressions, and measures what each compression step costs in operational accuracy; and partner with AI, autonomy, systems, and DevOps engineers to deliver production-ready edge builds.
WHAT WE ARE LOOKING FOR
Strong experience shipping ML models to production on resource-constrained hardware, with practical depth in quantization, pruning, distillation, and low-precision inference.
Strong grasp of the compute stack - GPU, NPU, and CPU performance, memory bandwidth, thermal and power limits, and profiling - and the ability to explain why a system is slow and fix it.
Hands-on experience with edge or embedded AI platforms and their inference toolchains (for example NVIDIA Jetson and TensorRT, ONNX Runtime, or comparable), and with real-time vision workloads such as detection, tracking, segmentation, and embeddings.
Strong Python and C++ (or Rust), with the ability to move between training frameworks and deployment runtimes and to design services that run reliably in a compact footprint.
Strong technical judgment, a measurement-first habit, and the ability to take substantial ownership and lead in a small senior engineering team.
Highly valued: experience compressing or serving LLMs and vision-language models on-device (small language models, KV-cache and memory tuning, speculative decoding), multi-object tracking or re-identification under real-time constraints, air-gapped or disconnected deployments, defense, public-safety, or government environments, and sensor integration (EO/IR, radar, passive RF).
BUILD WHAT MATTERS. GROW WITH CENTINUS.
At Centinus, the edge is where the product meets the customer's hardest constraints. You will help establish the foundation, make decisions with lasting impact, and expand your scope as the company grows.
Own the path that takes our AI from cloud development to standalone, disconnected systems operating beyond the lab.
Work directly with experienced leaders across AI, autonomy, systems, DevOps, and product, while defining the tools, benchmarks, and standards future teams will inherit.
Grow into broader technical and organizational leadership, including building the edge team, as the company scales.
See your work progress from engineering development to real operational use.
EQUAL OPPORTUNITY
Centinus is an equal-opportunity employer. We value exceptional ability, sound judgment, intellectual honesty, diverse perspectives, and people who take ownership of difficult problems.