Lev P.

I help founders turn product ideas into working software and help engineering teams build an AI-powered delivery process they can trust. I bring principal-level architecture, hands-on execution, and fractional CTO leadership across the software development lifecycle (SDLC). AI can accelerate implementation dramatically. The advantage comes from knowing what to build, giving agents clear direction, and proving that the result meets the business need. WHY THE CLAUDE COWORK STORY MATTERS Anthropic's Boris Cherny described Claude Cowork being built in roughly 10 days using Claude Code. That is a powerful example of how quickly an experienced team can build with AI. For your business, the opportunity is to combine that speed with explicit requirements, sound architecture, independent review, and validated releases. That is the focus of my SDLC Orchestration approach: one connected thread from discovery to delivery, with people accountable for the decisions. HOW I WORK 1. Discovery and shared intent Turn stakeholder conversations, transcripts, examples, constraints, and open questions into a shared understanding of the product. Make assumptions and unresolved decisions visible. 2. A living specification Define screens, actions, workflows, permissions, APIs, data changes, and edge cases. Track specification coverage: which intended behaviors are defined, implemented, and verified. 3. Tests before implementation Translate acceptance criteria into executable checks before building the behavior. Use unit, integration, and browser end-to-end tests to establish evidence that matters. 4. AI-assisted implementation and independent review Direct coding agents through bounded changes. Challenge the implementation with a different model and human engineering review for architecture, security, and maintainability. 5. QA against the actual build Validate the exact release candidate in a testing environment. Record acceptance and regression results, with engineering approval, QA acceptance, and production authorization treated as separate decisions. 6. Release, training, and feedback Ship the approved build with release notes and guidance. Feed real usage, support issues, and new discoveries back into the specification and tests. WAYS I CAN HELP - Fractional CTO leadership: technical strategy, roadmap, architecture decisions, engineering direction, and delivery oversight. - Product discovery and solution architecture: a reviewable specification, system design, priorities, and acceptance criteria. - AI SDLC adoption: practical workflows for coding agents, test-first development, independent review, CI/CD, and release validation. - Hands-on product delivery: move from idea or an existing codebase through implementation, QA, and a usable release. - Architecture and delivery reviews: identify gaps in requirements, permissions, integrations, testing, and operational readiness. TOOLING I AM DEVELOPING AT APP SCIENCES I am translating this methodology into a family of tools at different stages of development: - spec-dude — an exploration of a living map of product intent, connecting discovery to requirements, screens, workflows, APIs, and data. Its focus is specification coverage, ambiguity, and gaps teams can understand and resolve. - auth-dude — in development to make authorization policies, effective access, and the reasons behind permission decisions easier to inspect and review. - training-dude — an exploration of turning approved knowledge and software changes into role-specific lessons, knowledge checks, and walkthroughs. These initiatives reflect how I approach delivery: connect intent, implementation, evidence, and adoption. BACKGROUND I am a US-based software architect and fractional CTO with 20+ years of experience, including startup leadership and mission-critical initiatives at organizations such as the Federal Reserve and Fannie Mae. I work best with founders and teams who want to move quickly, stay lean, and make informed technical decisions. Tell me what you are building, where delivery is getting stuck, and what a successful outcome looks like. We can start with a focused consultation or scope a hands-on engagement through Upwork.