A specialized consultancy delivering production-grade AI systems, computer vision pipelines, and custom software — built by engineers who've done it at scale.
End-to-end AI and software engineering — from discovery through production deployment and ongoing support.
Custom CV model development for detection, segmentation, tracking, and classification. Built for real-world conditions, not just benchmarks.
Production ML infrastructure that actually scales. Training pipelines, model serving, and monitoring — done right.
Full-stack software built around your specific needs. APIs, dashboards, data pipelines, and internal tools — delivered clean and maintainable.
Scalable, reliable infrastructure for AI workloads. GPU cluster management, containerized deployments, and cloud architecture.
Technical leadership for teams navigating AI adoption. Architecture reviews, roadmap planning, and proof-of-concept builds.
A structured approach that keeps projects on track from first call to final deployment.
We dig into your problem, data, and constraints. You get a clear technical plan before a single line of code is written.
Iterative development with regular demos. No black boxes — you see progress weekly and provide feedback throughout.
Rigorous testing against real-world conditions. We don't ship until it performs reliably outside the lab.
Production deployment with monitoring, documentation, and optional ongoing support so you're never left stranded.
A selection of projects Charlie served as technical lead on, including publicly disclosed U.S. government SBIR contracts.
Technical lead on the Seahaven platform, a production AI system delivering synthetic data generation and computer vision model training for USAF Intelligence, Surveillance & Reconnaissance (ISR) missions. The platform powers autonomous identification of adversary systems in electro-optical sensor data, targeting Mean Average Precision >90%. Contract value: $1.25M (Phase II, 2024).
Technical lead developing an Automated Target Recognition computer vision model for integration on USAF aerial sensors conducting ISR and Guidance, Navigation & Control (GNC) at the edge. The model autonomously identifies adversary systems in electro-optical imagery with high precision, enabling real-time autonomous characterization without pilot intervention. Contract value: $75K (Phase I, 2025).
Technical lead on Spacehaven, a synthetic data generation platform for Space Domain Awareness and In-Space Servicing, Assembly & Manufacturing (ISAM). The system trains AI/CV models to support autonomous docking, debris remediation, and adversary characterization in the space domain using physics-accurate simulation. Contract value: $1.5M (Phase II STTR, 2023).
Technical lead on an Army-specific thermal synthetic data generation module and bespoke Thermal Object Detection Model. The system produces high-fidelity synthetic thermal sensor data integrated directly with the Army's MLOps and CV training pipelines, enabling autonomous object detection and tracking across varied operational environments. Contract value: $2M (Phase II, 2025).
Technical lead on a Phase I effort adding agent-based modeling to the Seahaven platform to enable autonomous prediction of adversary intentions and actions across multiple warfighting domains. The system leverages time-series synthetic data generation to train models that anticipate mobile adversary hardware and personnel behavior. Contract value: $75K (Phase I, 2023).
Software engineering on the Fusion platform — a state-of-the-art Live, Virtual & Constructive (LVC) content management and collaboration system for defense training. Fusion integrates AR/VR/MR, real-time GIS and IoT sensor data, and AAA-quality 3D visualization for scenario development, mission planning, rehearsal, and debriefing across DoD clients.
KC-Tech is led by Charlie Erway — a software engineer specializing in computer vision, machine learning systems, and production AI infrastructure.
Charlie brings deep hands-on expertise across the full ML stack: from training and fine-tuning CV models to deploying scalable inference systems on cloud infrastructure. His work spans Python, PyTorch, OpenCV, Docker, Kubernetes, and 3D rendering tools — applied to real problems, not toy datasets.
KC-Tech exists to close the gap between AI prototype and production. We work best with teams that have a real problem and want a technical partner who will own the outcome, not just the hours.
Proven, production-grade tools — chosen for reliability, not trend.
Tell us about your project. We'll respond with a clear technical plan — not a sales pitch.