Two decades turning high-stakes, FDA-regulated enterprise systems into working reality, now building local-first, sovereign AI that finally thinks the way my mind always has: automate, connect, simplify, and let the technology adapt to the person.
Long before I had the language, the tools, or the technical capacity to do anything with it, my brain was already working like an AI, always hunting for ways to automate, to standardize, to connect systems end to end, and to make something complicated feel simple and obvious to the person using it.
I've always preferred technology that adapts to people rather than forcing people to bend around the limits of a system. That instinct shapes how I see the full stack. To me it was never just the pragmatic delivery of features. The whole stack is a connected flow of human intentions, thoughts, emotions, interactions, and outcomes, and the technology's real job is to carry that flow as naturally as it can.
AI changed what's possible. Code is becoming a commodity, and vibe-coding environments now let me turn an idea into a working system at a speed that simply wasn't available to me before. The technology has finally caught up with how I always wanted to think and build, and I've spent the last stretch building relentlessly to prove it: hundreds of repositories, dozens of applications, whole platforms, each one a small argument that the gap between intention and working software is closing.
My own path mirrors that shift. I came up inside rigorous, compliance-driven worlds, pharma, life sciences, medical devices, where every system is validated, audited, and accountable. I've since moved into a far more open startup and open-source environment, with the freedom to experiment, build, and express these ideas directly. That's the connective thread of everything here: the discipline of structured, high-stakes systems, now married to the creative freedom and raw speed of AI-driven development.
Active SaaS platforms, discovery directories, analytics engines, and research portals running across the ReliaTrack AI-Labs network.
Turn raw intent, voice notes, and user stories into 21 CFR Part 11 compliant PRDs, SRS docs, and test matrices instantly.
Local-first entity discovery and semantic search over 65,000+ indexed tech stacks, libraries, and architectural benchmarks.
Multi-channel audience analytics, video transcript mining, and engagement intelligence for creator and brand growth.
Built for enterprises and startups looking to rapidly develop, stress-test, and deploy production AI MVPs and 24/7 autonomous agent systems.
End-to-end cloud infrastructure, IPaaS integration pipelines (n8n, ActivePieces, Make), CI/CD automation, and Proxmox/GPU cluster DevOps.
The sovereign R&D laboratory powering all research, GPU cluster nodes, and safety/eval frameworks.
Deep-dive research, architectural blueprints, and weekly digests on sovereign AI, multi-agent frameworks, and local LLMs.
The digital creative showcase and video demonstration hub for my vibe-coded AI Studio products and experimental prototypes.
It all runs out of ReliaTrack AI-Labs, the private R&D lab I founded to research and stress-test AI in depth, from local model infrastructure and multi-agent systems to the security and evaluation layers that make them trustworthy.
Epicurious is a local-first Agent Studio & Skill Forge I designed and built end to end: a control plane for creating, importing, security-scanning, and running AI agents, skills, and commands, with its own repo, one-command installer, and a harness that plugs into Claude Code, Codex, Cursor, Gemini, and the terminal, the same agents whether you drive them from a web dashboard or the CLI.



Underneath it sits 18 months of homelab engineering, hundreds of locally tested repositories spanning APIs, CLIs, full-stack apps, DevOps, and workflow automation across n8n, ActivePieces, Zapier, and Make.com. Epicurious began life as Anigar before I rebuilt it into the platform it is today, and the same lab has shipped production apps to real hosting, from Netlify to my own ReliaTrack domains.
Over 120+ applications built and engineered across AI Studio, IDEs, and local labs. Each is a distinct product with its own brand, UX, and business logic, evidence that the range is real and hands-on, not theoretical.
Local-first clinical intelligence on MedGemma, radiology second opinion, EHR omnisearch, and dermatology triage, specced with a full PRD and a self-hostable SurrealDB vector schema.
An enterprise career suite: live resume scoring, ATS keyword matching, a job-tracker CRM, interview prep with video, and one-click portfolio publishing.
AI code-component generation with an Express/Vite server, subscription tiers, PayPal/Stripe checkout, and deployment to Google Cloud Run.
Multi-agent workspaces with configurable personas, a draggable lab canvas, and super-agent/sub-agent dashboards, benchmarked against Vercel and Datadog.
An enterprise creative-direction tool with a full custom brand, an animated orbit UI, and a resilient Gemini image pipeline with backoff and rate-limit handling.
Browser and voice-driven coding environments, live-preview IDE plus speech-to-code with Drive integration, because I build the tools, not just use them.
A multi-agent command center: give the master agent (Aetherius Prime) a mission and it autonomously spins up and coordinates specialist sub-agents, legal analysis, code, creative writing, image generation, OCR, and audio, each on its best-fit model (Gemini 2.5 Flash, GPT-4o, Stable Audio). Ships with an engine that turns prompts into polished pages and decks.
Also spanning: OSINT & intelligence · legal e-discovery · sales & prospecting · enterprise ops & BI dashboards · project management with Gantt · HR & onboarding · fact-checking · creator analytics · media processing · directories · identity & email infra · education & games.
Before the AI work, I spent nearly two decades turning high-stakes, highly regulated mandates into seamless, high-adoption enterprise operations, at Teva Pharmaceuticals and Sparta Systems, with earlier roles at Ness Technologies and Mul-T-Lock.
I masterminded Teva's global SAP SuccessFactors LMS rollout under 21 CFR Part 11 validation, served as Business Owner for the ComplianceWire SaaS platform, and harmonized 12 disparate TrackWise quality-management instances into one global platform for critical FDA commitments, reaching ~92% adoption within six months and a ~47% reduction in compliance deviations.
Figures are self-reported and directional, drawn from program outcomes across regulated life-sciences deployments.
MBA, Technology Management & Information Systems, Tel Aviv University (Coller School of Management, Intel Future Leadership Program)
BSc, Industrial & Management Engineering, Information Systems & Knowledge Management, Technion, Israel Institute of Technology
Languages: English (full professional), Hebrew (native), Spanish (working)
Sovereign / local-first AI architecture · multi-agent orchestration · RAG & LLM evaluation · automated security gating · GxP & 21 CFR Part 11 validation · computer system validation · digital transformation · Lean / Six Sigma · master data management · P&L & global program leadership · executive stakeholder alignment.
Open to conversations on enterprise AI transformation, sovereign AI architecture, and executive advisory, globally.