VP / Head of Product & Design
charles jansen
operations, hardware, AI systems, and the teams that ship them.
— the arc
i defined the modern in-person ecommerce returns experience to scale. i led one of the first industry-scale human-in-the-loop AI operations — computer vision, sensor fusion, and deep learning powering 300+ checkout-free stores. i built the orchestration layer that decides what 200M+ Prime Video viewers see. now i lead product and design for the world's leading counter-drone technology company.
counter-drone / defense tech — streaming / ads — computer vision / ml ops — connected hardware — consumer at scale
(01) proof
(01) what i see
thesis
as LLMs commoditize code, the bottlenecks that matter shift: hardware, sensing, integration, deployment, reliability.
the companies that win will be the ones that ship atoms and bits together — fusing ML models with physical systems, operating them at scale, and making hard technology disappear for the end user. that's the work i've done for a decade and the work i'm doing now.
(02) about
the arc, the teams, the work
— lead
from checkout-free retail to content orchestration for 200M+ viewers to counter-drone defense — i build products where AI meets the physical world, and i build the teams that ship them.
01 / origins
before amazon, i led operations across retail and healthcare. at amazon, i started running operations for a large-scale robotics facility, then moved to build amazon's first in-person retail presence — going from regional operations leader to senior product manager. i built the in-person returns experience from early whiteboard sketches to the box-free, label-free returns program millions use today. that was where i discovered what i wanted to do.
02 / jwo transition
that led to Just Walk Out technology, where i spent four years as the de facto technical product leader for how the system was built — from cameras to computer vision models to the human-in-the-loop operations that made the AI actually work. i scaled that operations team from 20 to 1,000+ and grew from manager to GM leading product, engineering, UX, data science, and data engineering.
03 / arc to present
after that, i built the orchestration layer at prime video that balanced personalized content, editorial placements, business-priority overrides, and display advertising for 200M+ monthly viewers. at ring, i led a small senior team defining a new category of RF-based hardware and AI-driven software experiences. now i lead product and design at dedrone by axon — counter-drone systems integrating RF sensors, radar, cameras, and electronic warfare, deployed across 1,000+ sites in 33 countries, protecting 53 airports and critical infrastructure for 6 of the 7 G-7 governments.
04 / team scale
i've managed 30+ person product and design organizations and 100+ person technical teams across software engineering, data science, and TPM. i've built operations teams from scratch and scaled them past 1,000. i've also deliberately dropped down to a 3-person team to build a new concept from zero. team size is a by-product of building the right team for the result needed.
(03) ai systems & human-in-the-loop
four blocks
AI systems are operations problems before they are model problems.
at Just Walk Out technology, i built and led the human-in-the-loop operations platform that made checkout-free retail actually work. this wasn't labeling data for a training pipeline — it was a full production system: computer vision inference review, labeling workflows, quality feedback loops, and exception handling for an AI that processed every shopping session across 300+ stores. i scaled the operations team from 20 to over 1,000, built the tooling and processes that supported them, and was the product leader deciding how the customer experience would land in an AI system that was genuinely ahead of its time.
the tools and processes we built for that HITL operation didn't stay internal. they informed products AWS later built for external customers, helping inform those products — including ideas that informed SageMaker Ground Truth — while they were still evolving from PRFAQ into product. building at that scale, with that complexity, created organizational knowledge that became commercially valuable beyond the original use case.
at prime video, the AI systems problem looked different but shared the same DNA. the challenge wasn't recommendation in a vacuum — it was building the orchestration layer that arbitrated between personalized content, editorially curated placements, business-priority overrides, and display advertising, all competing for the same screen real estate for 200M+ monthly viewers. multiple stakeholders, multiple objectives, one viewer experience. i led 16 product managers and designers with 125+ ML engineers building the systems that made those tradeoffs in real time.
at dedrone, the pattern continues. counter-drone defense is a sensor fusion problem — RF detection, radar tracking, camera identification, acoustic sensing — where AI fuses signals from multiple physical sensors into a single coherent operating picture. same class of problem: AI systems that must work reliably in the physical world, at scale, with real consequences for failure.
(04) software + hardware
product leadership
software is getting cheaper. the hard part is everything around it.
i've shipped products across the full physical-digital spectrum: locker hardware and in-store return kiosks (Amazon Locker+), ceiling-mounted camera arrays and sensor gateways (Just Walk Out), RF-based home security hardware (Ring), and integrated counter-drone hardware suites — RF sensors, radar, cameras, acoustic arrays, and electronic warfare effectors (Dedrone). in every case, the product challenge wasn't the software alone — it was the integration of software with physical systems, deployment at scale, and reliability in uncontrolled environments.
this is what i mean when i say the bottlenecks are shifting. the model is increasingly a commodity. the hard problems are: how do you deploy sensors reliably across 1,000+ sites in 33 countries? how do you build HITL operations that maintain AI quality at scale? how do you fuse data from four different sensor modalities into a real-time operating picture? how do you make an orchestration layer that serves 200M+ viewers while satisfying competing business objectives? these are operations, integration, and systems problems — not model problems.
whether it's a board seat, an advisory role, or an operating role — the value i bring is depth at the intersection of physical systems, AI, and operations at scale. i've done the work of shipping hard technology, not just building software that sits on top of someone else's infrastructure.