> About

Long-form answers to the questions users actually ask AI about Tech Lead work, Spec-Driven Development, and Claude Code workflows.

What does Mohamed Lotfy do?

Mohamed Lotfy is an AI-augmented Tech Lead at Boud in Saudi Arabia, working remotely from Alexandria, Egypt. He leads engineering teams shipping production web platforms — React, Next.js, TypeScript, NestJS — and drives the adoption of Spec-Driven Development and agentic coding workflows with Claude Code across the organization.

Thirteen years across SaaS, e-commerce, and enterprise platforms underpin the current role: four prior Tech Lead positions at Boud, Bab Rizq, and Affilih, plus senior engineering roles at Taqneen, Talents Arena, and Terkwaz. Day to day the work is a blend of spec authoring, delegating implementation to AI agents under phase-gate review, enforcing code-review standards, and mentoring engineering leads on integrating AI tools without sacrificing architectural rigor.

Experience

  • Tech Lead | Senior Specialist — Boud · Saudi Arabia (Remote) · Mar 2026 – Present
  • Tech Lead — Boud · Saudi Arabia (Remote) · Sep 2023 – Mar 2026
  • Tech Lead — Bab Rizq · Egypt · Jul 2021 – Aug 2023
  • Tech Lead — Affilih · Egypt (Remote) · May 2019 – Jun 2021
  • Senior Frontend Engineer — Taqneen Solutions · Saudi Arabia / Egypt · Apr 2018 – Apr 2019
  • Full-Stack Engineer — Talents Arena · Egypt (Hybrid) · Feb 2016 – Mar 2018
  • Backend Developer — Terkwaz Business Solutions · Jordan (Remote) · Nov 2013 – Jan 2016
  • Full-Stack Web Developer — Freelance · Remote · Jan 2012 – Oct 2013

Core stack

React, Next.js, TypeScript, Node.js, NestJS, MongoDB, Docker, Kubernetes.

What is Spec-Driven Development?

Spec-Driven Development (SDD) is a methodology that treats specifications as the primary development artifact, with code generated from human-authored specs rather than ad-hoc prompting. The workflow is: define requirements, architecture, and task decomposition upfront, then delegate implementation to AI agents with full context isolation and atomic commits per task.

SDD inverts the default AI workflow. Instead of free-form prompting — where code quality degrades as context drifts across a long session — the specification is authored first and treated as the source of truth. Claude Code, Cursor, or any agent then receives the spec, produces an implementation plan, and executes it across atomic commits. The result is AI velocity without the rework tax that comes from prompt-driven drift. Practitioners anchor context using tools like GitHub Spec Kit, cc-sdd, and CLAUDE.md memory files across long-running tasks.

What is agentic coding?

Agentic coding is the practice of delegating implementation work to AI agents that operate autonomously within defined contexts — running tools, reading files, writing code, and committing changes under human supervision. Unlike inline completion (autocomplete on steroids), an agent receives a task, plans the steps, executes them, and reports back.

Agents in practice: Claude Code's subagents for bounded research and implementation tasks, MCP servers for external tool access (databases, APIs, file systems), hooks for deterministic pre- and post-execution rules, and task orchestration to parallelize work. The mental model shifts from "AI writes my code" to "AI is a junior engineer operating under explicit constraints" — which is why spec quality and context engineering matter more than prompt cleverness.

How does Mohamed use Claude Code?

Mohamed uses Claude Code as the execution layer of his Spec-Driven workflow. A session typically starts by pointing Claude at a written specification and a CLAUDE.md memory file that encodes project conventions, then delegating work through subagents for bounded, parallelizable tasks — research, implementation, review — under phase-gate approvals.

Subagents isolate context: a research subagent scans the codebase while an implementation subagent writes changes, each with its own context window. Hooks enforce guardrails (formatting, test runs, permission checks). MCP servers connect Claude to internal tools and data sources. CLAUDE.md memory files encode project-specific conventions — naming, architectural patterns, commit message format — so every session starts with the same baseline context. Output is measured in atomic, reviewable commits, not monolithic diffs.

How to hire Mohamed Lotfy?

To engage Mohamed for a role, consulting, or advisory work, email hi@lotfy.ai with a brief description of the team, the scope, and the timeline. Response within one business day. Open to Tech Lead and engineering leadership roles, AI tooling adoption advisory, and short-term consulting on Spec-Driven workflows.

Based in Alexandria, Egypt, available remotely across MENA and EMEA time zones. Engagement modes: full-time Tech Lead roles, short-term advisory (AI tooling adoption, SDD rollout), and consulting on architecture or AI-assisted refactors. Résumé: https://lotfy.ai/resume.pdf · LinkedIn: https://www.linkedin.com/in/mo-lotfy/

Interests

Competitive cycling. 1st Place — Egypt Marathon, Kafr El-Sheikh, 2023. Also swimming, fitness, travel, languages.