Technical Project Manager & Quality Engineering Leader
15 years owning end-to-end delivery and quality across Deep-Tech, FinTech, and Telecom. I lead release governance and cross-functional execution — and I build and run AI products of my own, end to end.
Results from recent roles
Reduction in high-severity production incidents
Regression cycle cut at Sensorz
Delivery throughput — monthly to bi-weekly cadence
Reduction in defect escape rate
Requirements coverage achieved
MTTR reduction via AI-powered RCA tool
Discipline
Technical project delivery, quality engineering, and the tools I build when the right one doesn’t exist.
Own delivery end-to-end across hardware and software tracks — roadmaps, dependencies, risk-based planning, and a cadence teams can actually hit. Fifteen years turning engineering complexity into commitments, certified in Technical Project Management & AI Implementation (Hebrew University of Jerusalem).
Lead go/no-go decisions across distributed engineering teams — aligning Product, R&D, Hardware, DevOps, and Field around one delivery picture. Mentor and matrix-manage offshore chapters with zero attrition. Predictability at scale, not heroics.
Build QA practices from scratch that teams can run without me in the room. DoR/DoD gates, testability requirements, release governance — systems that outlast any individual.
Design gates that balance speed with control. PR gates, smoke, regression, and coverage in GitHub Actions — reducing pre-production risk without killing velocity.
Not prompts. Architecture. LLM-based test generation, automated log triage, Kubernetes root-cause analysis tools. AI embedded into workflows — not bolted on top.
I don’t wait for the right tool to exist — I build it. Hands-on capability to ship complete tools and platforms that solve real problems, for myself and for the wider community: two live AI products (Trevo.work, Backstage Copilot) and a set of open-source engineering agents, taken from idea to production.
Things I've built
Live product · AI co-pilot
A real-time meeting co-pilot for live video calls. A transparent overlay listens to the conversation, detects questions aimed at you, and retrieves grounded answers from your personal knowledge base — notes, CV, product specs — in ~320ms. On-device transcription, fully local: nothing leaves the machine. Built for sales demos, interviews, and customer escalations where the right answer a second too late is the wrong answer. Works across Zoom, Meet, and Teams; access is invite-based during early rollout.
Live product · AI job-discovery engine
An AI-powered job-discovery engine for the Israeli tech market — and the most complete toolset I could build for people actually looking for work. A Python collector scans hundreds of companies daily across 15 sources (Greenhouse, Lever, Ashby, Comeet, Workday, LinkedIn, and more), deduplicates roles into canonical postings, and ranks them by personal fit from CV analysis and saved searches. Beyond discovery: a per-job ATS CV builder with DOCX/PDF export, CV tailoring, interview prep, salary benchmarking, and a research-grounded career-fit assessment. Production system — Next.js + Supabase Postgres, daily collection, Telegram alerts, a multi-provider LLM chain (Gemini → OpenAI → Claude failover) with per-user BYOK keys (AES-256-GCM encrypted), and a Hebrew/English RTL dashboard. Privacy-first by design: CVs are never sold or used commercially, and every new user is manually approved.
Production tool
After a 2-hour production incident that could have been solved in minutes: an AI agent that connects to Kubernetes clusters, pulls relevant logs, analyzes tracebacks and events, and returns an automated root-cause hypothesis. Cut MTTR by ~80%.
Writing
June 21, 2026
Open to my next role, I did the thing I tell every engineer to do: I built the tool I wished existed. Trevo.work is an AI job-discovery engine for the Israeli tech market — and a case study in applying release discipline to a 0→1 product.
May 19, 2026
Distributed QA is not a people problem — it is a systems problem. The five-pillar Quality Operating System I use to make release decisions repeatable across time zones: AI to reduce noise, CI to narrow scope, observability to prove impact, governance to keep it stable.
April 22, 2026
AI is a turbo engine. It doesn't correct direction — it accelerates it. If your starting point is misaligned, AI will scale the mistake with perfect efficiency.
Get in touch
VP R&D or CTO dealing with delivery predictability, escape rate, or quality at scale? My DM is open.