25 years scaling R&D orgs — now proving the AI-era playbook hands-on. I've led multi-team engineering groups through hypergrowth, incidents, and global delivery. Today I pair that judgment with AI-First methods to ship at a velocity most teams can't match.
The 1Month apps were the pressure test. The next step is bringing this playbook to a full R&D organization — I'm looking for my next engineering leadership role.
Two decades of engineering leadership — measured in teams grown, users served, and outcomes delivered.
Grew and led 15+ engineering teams — organic, near-shore and offshore, distributed globally — across Kaltura, Arbox, Assure, and FTS.
Scaled the R&D org behind a B2B2C platform serving 500,000+ active users, re-architecting AWS for high availability under heavy traffic.
Initiated and led an automation initiative, automating complex workflows for hundreds of thousands of dollars in operational savings. Integrated AI via RAG on AWS Bedrock.
Delivered enterprise software to 20+ Tier-1 clients across the US and Europe under strict compliance and security standards.
The playbook I bring to engineering teams: architectural judgment paired with AI-First execution — accelerating delivery while holding the line on security, Business ROI, maintainability and observability.
Harnessing AI to accelerate delivery. Emphasizing prompt engineering, skills, harnesses, and cross-model prompt testing. Moving from AI-First → AI-Only, achieving incomparable velocity.
Designing robust microservices, real-time spatial pipelines, and offline-first edge applications built on reliable, enterprise-grade cloud foundations.
Aligning technical strategy with commercial goals. Prioritizing features that drive business metrics, minimize technical debt, and ensure long-term system health.
Implementing OAuth, strict JWT RBAC, biometric guards, and automated PII redaction. Building security compliance natively into the architectural blueprint.
Deep architectural judgment meets AI-First methodology at 10× velocity.
I went hands-on by design — to lead AI-era engineering from fluency, not nostalgia. The 1Month apps are how I pressure-tested the methodology I now bring to teams.
I'm a tech expert with 25 years deep in engineering. Today I build solo, leveraging AI for 10× velocity — shipping production apps single-handedly that would normally take a small team months.
Through 1Month my first app went from an idea to working on my mobile device in one month. Next ones were significantly faster — backed by production CI/CD, cloud-native architecture, and ready to scale. No prototypes. No demos. Real apps, real complexity.
Behind the velocity is two decades of architectural leadership: scaling global R&D teams across Kaltura, Arbox, Assure, and FTS; leading through leaders — growing team leads into autonomous managers with 100% retention while more than doubling the group; calmly managing incident response; aligning technical strategy with commercial ROI. The same judgment that shaped enterprise platforms now shapes every line of code, every infrastructure decision, every release.
Proving the exponential velocity of AI-First methodology: the first app took 1 month from idea to working product on my device; The second, utilizing established prompts, took a single night. In under 6 months, single-handedly engineered and shipped 7 applications across iOS, Android, watchOS, macOS, and Web — backed by CI/CD pipelines with automated and load testing, running on Cloud infrastructure. Every mobile app was delivered with uncompromising enterprise-grade architecture and rigorous security standards.
Direct leadership of a multi-disciplinary engineering group (4 teams) supporting the company's primary growth engine in a new business line. Served as central technical authority aligning R&D delivery with commercial goals. Initiated and led the development of a strategic product built on Node.js and React, later extended with complementary AI capabilities on AWS (Bedrock) to automate complex workflows for hundreds of thousands of dollars in operational savings.
Led 3 teams (organic, outsource & offshore) during growth, supporting B2B2C customers and 500,000+ active users. Built scalable R&D organizational structure to fit aggressive business goals. Overhauled the AWS cloud architecture to ensure high-availability for a high-traffic PHP/Angular/React platform, and implemented metrics-driven engineering KPIs.
Owned the full technology lifecycle from early-stage infrastructure to global deployments. Scaled data-intensive BI solutions for Testing & Development domains digesting millions of records daily. Managed enterprise delivery for 20+ Tier-1 clients across the US and Europe, maintaining strict compliance/security standards.
Managed 5 teams (organic, near-shore & offshore) overseeing complex CRM solutions for the telecom sector. Championed the group shift to Agile, drastically cutting development cycles and fostering cross-departmental collaboration.
Progressed from core developer to Team Leader across multiple business domains (Retalix, NTR, Casdim).
Velocity without amnesia. The 1Month portfolio runs on an operating system for AI-First work — a version-controlled company brain, reusable skills, production agents, and adversarial cross-model review — so every AI session starts with full context and ends by writing back what changed.
A git-versioned memory that runs the whole portfolio — the single source of truth for everything the code can't tell you: naming traps, account ownership, decisions with their rationale, dead ends proven once so no AI session re-burns the time. Every work session opens by reading it and closes by writing back what changed. Its rule: brain pages are secondary sources — primary evidence (git history, CI logs, store consoles) always wins. Private by design (client and account context lives there); the numbers are the tour:
Four version-controlled skills — security audit, security remediation, product-manager audit, and the brain's own update protocol — written once, symlinked into every project, and kept project-aware through per-repo context files.
An autonomous agent lives on Cloud Run serving B4UGO: first-login-of-day alerts delivered to Telegram over HMAC-signed webhooks. Not a demo — live in production and verified end-to-end.
One model's plan is a hypothesis, not a verdict — rival models poke holes in each other's plans before a line of code ships. Adversarial review as a standing rule, not an occasional trick.
Every production repo carries an AI-facing architecture doc, dated audit trails, and handover notes recording what was verified versus assumed — so a fresh session is productive immediately.
Every public number on this site sits in a dated manifest beside the command that verified it. Claims decay — the manifest catches drift by re-running the evidence, not by luck.
Four phases, 13 rules for leading AI-First engineering — published on LinkedIn as it happened, wins and dead ends included.
The leadership is the headline — this is the proof. Seven production apps, solo-built with AI-First methods: the playbook above, pressure-tested in code.
Location-based shopping reminders. Add items to your list, associate them with stores, and get notifications when you're nearby.
Civic issue reporting for municipalities. Citizens report geolocated problems; city employees triage and resolve via dedicated admin portals.
Private, offline proximity alerts. Import any Google My Map, then get a local notification when you're near a saved pin - even with the app closed. No backend, no accounts.
Family expense tracker - log expenses in 3 seconds, share a household dashboard, and track budgets with Dynamic Island live updates.
Personal inventory manager - snap a photo, tag to a storage location, find your tools and gear in seconds. 100% offline, zero cloud.
Offline EXIF extractor for photographers - scan massive RAW folders, format camera settings for social media, batch export with privacy stripping.
Smart duplicate file finder - scans folders using content hashing or weighted name similarity, helps you reclaim disk space.
Every app uses production-grade architecture, security, and operational patterns.
Real deployments, real domains, real CI/CD - not localhost demos.
Building fast doesn't mean cutting corners. Every platform has automated tests in CI, load testing scripts, real-time error tracking, and uptime monitoring.
882 automated tests across the portfolio, all passing — re-verified by running
every suite on Aug 3, 2026. On B4UGO, CityFix and Tap-Ching, every push to main runs the
suite; B4UGO's Cloud Run deploys fire only on a green run.
k6 scripts validate that the B4UGO serverless backend handles real-world traffic patterns without degradation.
tests/load/ — part of the monorepo, versioned alongside services
How I monitor my production apps without a team — custom-built admin dashboards with real-time health checks, error tracking, analytics, and CI/CD visibility. Every platform has its own System Admin portal.
@sentry/nestjs on 3 backends + @sentry/react-native on mobilescan_triggered, app_open, item_added,
settings_changed
Dispatches from the 1Month experiment — posted on LinkedIn as it happened, wins and dead ends included.
“The risk for a senior engineering leader isn’t falling behind on strategy. It’s quietly losing touch with how the work actually gets done — and starting to lead from nostalgia instead of fluency.”
So I built. Seven production apps in under six months, solo, with AI — across iOS, Android, watchOS, macOS, and web.
“AI amplifies a sharp spec or a vague one at the same speed. The spec is your real IP now.”
Part 1 of the manager’s AI playbook — set the foundation: explore before you decide, make clarity the multiplier, keep architecture human.
“Take one model’s plan to a rival and make it poke holes. ‘Claude Code says do X’ is a hypothesis, not a verdict.”
Part 2 — run AI as a system: match the model to the task, pit models against each other, run a token economy, layer your instructions.
“The scarce skill is no longer typing the code. It’s judging what came back.”
Part 3 — hold the quality line: raise the review bar, own hardening explicitly (AI ships the happy path), and institutionalize audits — quality as a standing ritual, not a one-time gate.
“A ranking is a decision. A list is not.”
Part 4, the finale — ship and decide: force-rank with AI, instrument and let real usage set priorities, and lead the “no.” The whole playbook: four phases, 13 rules.
“I built my projects a brain. Its first act was calling me a liar. (Technically, it called my documentation a liar.)”
How a 12-project solo portfolio got a shared, git-versioned memory — and immediately caught confident falsehoods the docs had been telling for months. The rule that emerged: primary evidence wins; docs are just claims.