Mostafa SakhiriMostafa Sakhiri · Frontend Software Engineer

Engineering Impact

Engineering Philosophy

How I think about building software

I believe engineering is not about shipping individual features. It is about continuously increasing an organization's ability to build software.
  • Systems over features

    I optimize for the system that produces features, not the feature itself. A well-designed system makes every future feature cheaper to build than the last.

  • Reduce complexity over time

    Complexity compounds silently until it stalls a team. Every architectural decision should leave the codebase easier to reason about than it found it.

  • Build reusable foundations

    Shared abstractions, once built well, pay for themselves on every feature that follows. I invest in foundations before I invest in one-off implementations.

  • Developer Experience compounds over time

    A faster feedback loop today becomes faster delivery for years. DX is not a nice-to-have — it's a multiplier on everything the team ships afterward.

  • Performance should be built into the platform

    Performance should be a property of the platform, not a follow-up optimization pass. When it's inherited by default, every new feature starts fast.

  • AI should augment engineering, not replace it

    Coding agents are only as good as the context and guardrails around them. I design the workflows, documentation and review processes that let AI raise a team's throughput without lowering its bar.

  • Every architectural decision is a business decision

    Architecture shapes delivery speed, maintenance cost and how long a platform stays adaptable. I make those trade-offs explicit to the business, not just the codebase.

See the full framework →
Engineering Impact

Four contributions that changed how a team builds software

Each one reduced the cost of building the next feature — not just this one.

Accelerating Engineering Delivery

Tripled frontend engineering throughput by designing and deploying an AI-assisted development system enabling developers and coding agents to collaborate efficiently while maintaining high engineering quality.

frontend engineering throughput
  • Selected by engineering leadership to lead a three-month initiative dedicated to improving engineering productivity through AI.
  • Introduced Claude Code and standardized AI-assisted engineering workflows across the frontend team.
  • Built reusable engineering documentation, Context Engineering practices and reusable AI skills to provide coding agents with the necessary project knowledge.
  • Automated technical and functional Pull Request reviews before human review to reduce QA feedback and improve delivery quality.
Claude CodeContext EngineeringAgentic EngineeringGitHubGitHub ActionsTypeScriptReact
Read the full story →

Scaling Frontend Delivery

Reduced implementation time for new product pages by 50–70% by transforming the Design System into an engineering platform powering nearly 80% of the new React application.

50–70%faster implementation
~80%platform adoption
  • Owned the Design System used by a frontend team of 5–8 engineers.
  • Designed a reusable DataGrid integrating data fetching, virtualization, deferred rendering and common business interactions.
  • Embedded accessibility, UX consistency and performance directly into reusable engineering foundations.
  • Standardized reusable abstractions shared across multiple applications and libraries.
ReactTypeScriptStorybookAG GridDesign TokensTailwind CSSRadix UINx
Read the full story →

Making Performance the Default

Reduced Largest Contentful Paint by three times on the platform's most frequently used pages by embedding performance optimizations directly into reusable frontend foundations.

LCP improvement
  • Integrated virtualization, lazy rendering and intelligent data loading into shared platform components.
  • Eliminated unnecessary rendering and redundant network requests across complex user interfaces.
  • Investigated production incidents and delivered targeted hotfixes to continuously improve platform stability.
ReactTypeScriptTanStack QueryVirtualizationLazy LoadingPerformance ProfilingSentryAmazon CloudWatch
Read the full story →

Evolving the Frontend Architecture

Helped evolve Concord's frontend platform from a legacy Ember application toward a scalable React architecture designed for long-term maintainability.

  • Participated in the migration from Ember.js to React.
  • Designed modular architecture and reusable abstractions supporting long-term evolution.
  • Reduced technical debt through continuous architectural improvements.
  • Collaborated with Product, Design and Engineering teams to guide technical decisions.
ReactTypeScriptNxArchitectureDesign System
Read the full story →

Building an AI-Native Product Experience

Designed and shipped a conversational AI layer inside the platform that lets users fetch exactly the data they need, create new resources, and chat directly with an individual resource such as a contract — replacing multi-step forms and dashboard navigation with natural language.

  • Built natural-language data retrieval so users can ask for the exact record or data point they need instead of hunting through dashboards and filters.
  • Designed a conversational resource-creation flow, letting users generate structured resources like contracts through dialogue instead of long-form UI.
  • Implemented resource-scoped chat, letting users open a specific record and ask grounded questions about its actual content rather than generic AI answers.
  • Combined tool-calling and structured data access to keep every response grounded in the platform's real data — the same grounding-over-hallucination discipline behind this portfolio's own assistant.
AI-assisted EngineeringAgentic EngineeringLLM Tool CallingCopilotKitReactTypeScript
Read the full story →
How I Think

The models behind my engineering decisions

Feature by featureSystems thinking

Systems compound. Features don't.

A one-off feature solves one problem. A system solves the next ten. Systems thinking is slower in week one and faster than the alternative by week eight — and the gap keeps widening.

FasterfeedbackHigherconfidenceMoreshippingNewsignal

Developer Experience is a flywheel, not a checkbox.

Better tooling reduces friction. Less friction increases the rate of shipping. More shipping surfaces the next tooling gap sooner. I invest here first because it accelerates everything built after it.

BeforeAfter

AI multiplies engineering judgment. It doesn't replace it.

The teams that get the most from AI-assisted development have the clearest context, the strongest guardrails and the highest existing engineering bar. AI amplified throughput 3× because the review process, documentation and skills around it were built first.

See the full framework →
Featured Engineering Story

Tripling Engineering Throughput through AI-Assisted Engineering

Why I was trusted to lead a three-month initiative that tripled how fast the frontend team could ship — and how AI became a reliable part of the engineering workflow, not a shortcut around it.

Read the full story
frontend engineering throughput
Technical Stack

The tools behind the systems

JavaScriptReactTypeScriptNext.jsTanStack QueryTanStack RouterViteNxEmber.jsSingle-Page ApplicationsStorybookDesign TokensRadix UITailwind CSSAG GridFigmaZodReact Hook FormAI-assisted EngineeringAgentic EngineeringContext EngineeringClaude CodeAI ElementsPlaywrightReact Testing LibraryJestPercyAWSDockerGitHub ActionsGitLab CISentryAmazon CloudWatchSQLMetabase
Engineering Journal

Latest from the journal

Read the articleAugust 7, 2026

Most Engineering Decisions Aren't About Technology

A framework for reasoning about trade-offs, architecture and impact — the eight principles behind every technical decision I make, and why the technology is rarely the interesting part.