Full-Stack Engineering · Open Q3 2026 · Los Angeles
Production software for complicated real-world work.
I'm Yakir Yaacov, a full-stack engineer who designs and ships operations software, AI-assisted workflows, and TypeScript products. From claims and document systems to contract automation and commerce, this portfolio explains what I built, how it works, and the tradeoffs behind it.
Type / to see all commands · click ./check_availability.sh to run
>
Available commands
/help — list all commands
/projects — all projects
/writing — engineering notes
/stack — full tech stack
/background — about & approach
/work-experience — career timeline
/availability — current status
/contact — get in touch
/clear — clear terminal
0+yrProduction Engineering
0+DB Models Shipped
0+API Endpoints Built
0+Async Tasks Deployed
0+3rd-Party APIs Wired
About
⌨️
Production-Driven Architecture
Every framework, database choice, and optimization pattern I use was learned while shipping real software under real production constraints. My credentials aren't found in computer science textbooks — they are live in production.
🏗️
Systems-Level AI Integration
My engineering foundation predates production LLM tooling — which means I understand precisely where AI integration breaks: context-window failures, token waste, non-deterministic pipeline output. I build the scaffolding that makes AI reliable at production scale.
⚡
Extreme Ownership. High Velocity.
I operate like a senior engineer embedded in a startup: I take the spec, make the architectural calls, and ship the system — no committee required. Founders hire me when they can't afford to babysit a developer or wait three months for an MVP. I'm the single-hire that replaces a team.
🌍
Los Angeles Based · Available Worldwide
I work remotely with founders, startups, and technical teams across the US, UK, and beyond. Available for both project-based contracts and ongoing fractional engineering roles. Timezone: PT — flexible for critical builds.
Timeline
2020—NowSenior Software Engineer II · All Phase Consulting. Sole senior engineer behind Claimet App: a production, multi-tenant Django platform for claims, restoration operations, AI document workflows, reporting, and field collaboration.
2019—2026Senior Full-Stack Engineer · Independent. Delivered production systems across consumer social, Shopify Hydrogen commerce, fintech, Web3, AI automation, and TypeScript products.
2018—2019Backend / Systems Engineer · Confidential Layer-1 Startup. Built C++ node infrastructure and Rust cryptography modules spanning P2P networking, validation, mempool behavior, consensus, and testnet operations.
2017—2021Founding Engineer · Secretariat. Designed and operated a Python crypto-trading automation platform for hundreds of paying users, including on-chain integrations, licensing, and continuous production operation.
2016—2017Full-Stack Web Developer · ImageItIn. Built production client sites and internal tools with PHP/Laravel, REST APIs, crypto products, and multi-brand deployment pipelines.
Operating Philosophy
Architecture decisions documented, not assumed
Code is written to be maintained, not just to ship
One round of questions. Then heads-down execution.
Production means tested, deployed, monitored
No spec work. No unpaid trials. Real projects only.
What I Take On
Senior operator. One hire. Full ownership.
I don't need a team, a project manager, or weekly check-ins. You hand me the spec and I ship the system. Here's what that covers.
Backend & Systems
Enterprise Applications
End-to-end production systems built for data volume, async load, and long-term maintainability. Architected from the ground up, not scaffolded from a tutorial.
Django & Node.js / Express REST APIs
PostgreSQL + MySQL database architecture
Celery + Redis async task pipelines
Multi-engine PDF generation (ReportLab, WeasyPrint)
Webhook ingestion + HMAC verification
Docker · Nginx · Linux VPS deployment
Frontend & Product
Full-Stack Interfaces
React and TypeScript frontends that aren't an afterthought. Tight component architecture and UI that doesn't embarrass a premium product.
React · Vue · Next.js · TypeScript · Vite
Component architecture + custom design systems
WebGL / Three.js immersive UI
Shopify storefront + ecommerce integrations
Mobile app development (React Native)
Performance-first — Lighthouse 100 targets
AI & Automation
Pipelines & Intelligence
Not ChatGPT wrappers. Real AI integrations wired into production workflows — vision models, concurrency pipelines, intelligent data processing at scale.
Claude API (Vision + Text) production integration
ThreadPoolExecutor concurrent inference pipelines
Automated data extraction + processing tooling
Solidity ERC-20 smart contracts (OpenZeppelin)
BTCPay Server + on-chain crypto payment systems
Intent-based NLP chatbot backends
Featured Enterprise Build
Enterprise SaaS · Insurance Tech · Django + AI
Water Mitigation Claims Platform
The Problem
Insurance adjusters managing water damage claims across multiple properties had no centralized system. Field data, sensor readings, scope documents, ALE lease agreements, and client billing all existed in disconnected spreadsheets — creating compliance risk and billing delays on every claim.
System Architecture
End-to-end Django 4.2 web application managing the full lifecycle of water damage claims — from property intake and drag-and-drop room building, through Encircle API webhook sync, Claude Vision AI field verification, Excel auto-population, ALE lease pipeline, and multi-engine PDF generation. Deployed for All Phase Consulting at claimetapp.com.
These projects include production systems, public demos, and focused product builds. Each case study covers the problem, implementation, and engineering decisions so you can evaluate more than a screenshot.
SaaS · AI Pipeline · Polyglot Architecture
TaskScribe
The Problem
Contractors waste hours transcribing meeting notes, manually extracting action items, and emailing clients with status. Every step is a context-switch away from billable work.
System Architecture
AI-powered note transcription → structured task extraction → automated client delivery. Built as a polyglot monorepo: Next.js 15 (App Router, TypeScript) owns the product and database; a Python FastAPI service owns the AI pipeline. Decoupled by a Redis queue with HMAC-signed callbacks. Open on GitHub:ya1-oa/taskscribe
LLM extraction: structured JSON schema output (no freeform)
HMAC-signed callback → web app persists + emails client
Job audit trail: every run is a traceable DB row
SaaS · React Flow · Contract Automation
ContractFlow
PROBLEM: Contractor agreements are assembled across disconnected emails, milestone sheets, payment notes, and static templates
Visual agreement workspace that turns parties, scope clauses, milestones, compensation, and delivery terms into a validated contract model. The interactive demo shows node-based planning, contract-readiness checks, and the path to signed PDF generation.
PROBLEM: Contractors lose scope clarity and billing disputes because project milestones live in email threads — never a shared, auditable system
Visual project management + dynamic contract generation SaaS. React Flow node canvas where milestone nodes connect into a project timeline. As nodes are created and wired, a contract drafts in the background. As milestones complete, the system auto-emails the client a progress report — no manual updates.
PROBLEM: Sales teams burning hours per day on manual prospecting, enrichment, and outreach that should be a fully automated pipeline
Automated B2B lead generation infrastructure in Python. Multi-source prospect discovery, data enrichment against external APIs, qualification scoring, and sequenced outreach — from raw domain list to scored, enriched lead in CRM with zero manual steps in the pipeline.
PROBLEM: On-chain referral payouts required zero manual intervention at scale
Gamified Web3 metaverse platform with a production referral economy. The Villager model handles crypto wallet registration, LTC commission auto-payouts via BTCPay pull-payments, and HMAC-verified webhook processing — every payout triggered automatically, zero human touchpoints required.
PROBLEM: Field techs manually renaming hundreds of sensor photos — hours of dead time per job
Python desktop tool using Claude Vision AI to batch-read and rename field sensor images. 4+ concurrent API threads via ThreadPoolExecutor, real-time progress tracking, manual correction panel for unreadable screens, and per-subfolder ZIP export. What took hours takes minutes.
4xConcurrent Threads
100%Auto-Processed
PythonClaude Opus 4ThreadPoolExecutorTkinter GUIPillow
PROBLEM: No accessible tool for aspiring quant traders to backtest strategies without institutional-grade platforms
Full-stack algorithmic trading education and simulation platform. Interactive lab for backtesting strategies, paper trading, and technical analysis — TypeScript frontend with a production Node.js backend and persistent user data in PostgreSQL.
PROBLEM: A premium botanical brand needed an editorial storefront that could demo independently and switch to live Shopify data without a frontend rewrite
Production-ready green botanical storefront built with Remix, React, TypeScript, and Hydrogen-style commerce patterns. It ships with a seeded product catalog, search, collections, variants, persistent cart behavior, SSR, and a clean Storefront API adapter for a future Shopify catalog.
PROBLEM: Gig drivers see offer totals but lack immediate insight into mileage, time, zone performance, and true hourly earnings
Local-first driver operations app for evaluating offers, tracking work sessions, comparing zones, and measuring actual earnings. Designed as a mobile-ready PWA with offline behavior and a fast decision workflow.
PROBLEM: Referral commissions paid manually — error-prone, slow, and impossible to scale
Production referral commission system on BTCPay Server pull-payments. Auto-creates and caches a shared commission pool, processes on-chain LTC payouts with autoApproveClaims, handles API error codes (overdraft, amount-too-low) with full recovery logic. Zero-touch payout infrastructure.
PROBLEM: Emotional support app needed structured conversation routing without per-token LLM costs at scale
Conversational AI backend for emotional support and personal growth coaching. Deterministic rule-based router using regex-pattern matching against a structured intents corpus — eliminates LLM latency and runtime costs for interactions where response structure is fully predictable. Dream journaling, affirmations, goal-setting with MySQL persistence.
PROBLEM: Restaurants needed QR menus that convert to loyalty — not just static PDF replacements
QR-led guest engagement platform that transforms digital menus into a loyalty engine. Guests scan, browse, order, and accumulate rewards — operators get analytics on item performance and repeat visit rates.
I understand systems at the architecture level, not just the API call. Every tool below has been used in production code, not a Hello World tutorial.
Backend & Data
Django 4.x · Python
Node.js · Express.js
PostgreSQL · MySQL
Celery · Redis
REST API design + auth
WebSockets · SSE
Frontend & UI
React · Next.js · Vite
TypeScript · ES6+
Vue.js
Three.js · WebGL
React Native (Mobile)
Shopify Liquid
AI & Blockchain
Claude API (Vision + Text)
ThreadPoolExecutor pipelines
OpenAI API integration
Solidity ^0.8 · ERC-20
OpenZeppelin · Hardhat
BTCPay Server API v1
Infra & DevOps
Docker · Docker Compose
Nginx · Gunicorn
Linux VPS · DigitalOcean
GitHub Actions · CI/CD
SMTP · Pixel tracking
Webhook + HMAC security
How I Operate
Brief in. System out. No hand-holding.
01
Project Brief
You submit a brief with the build scope, constraints, and budget. I read it once, ask one round of clarifying questions, and that's it. No weekly discovery calls needed.
02
Architecture & Proposal
I deliver a scoped technical proposal — stack choices, architecture decisions, delivery timeline, and fixed-price quote. You get it within 48 hours. No vague "let's explore" responses.
03
Build in Isolation
I go heads-down and ship. You get progress updates at agreed milestones — not daily standups. I operate with extreme ownership. Your critical path isn't held up waiting on me.
04
Deploy & Hand Off
Production deployment, documentation, and a clean handoff. Systems I ship are maintainable. You're not buying a black box you can't debug six months later.
Client Reviews
Straight from the clients.
No agency-curated quotes. Verified reviews from founders who hired me to ship critical systems — not redesign a landing page.
"
I've hired developers who needed constant direction, daily standups, and still delivered late. This was not that. I handed over a high-level brief for our claims management platform — field sensor automation, ALE pipelines, third-party API sync, multi-engine PDF generation — and came back to a fully architected, deployed system. No check-ins. No excuses. No scope creep. The speed and level of autonomous execution was unlike anything I'd seen outside of a senior engineering team. If you need a developer you don't have to babysit, this is the only call you need to make.
J
JosephFounder · All Phase Consulting · ClaiMetApp
Verified Client · Authenticated Portal
"Dropped a Solidity spec on a Friday. Had two gas-optimized ERC-20 contracts with full Hardhat coverage by Monday. Zero back-and-forth. Exactly what I asked for."
— Anonymous Client · Web3 Platform · Verified
Client Review Portal
All testimonials are submitted through a private authenticated intake. No anonymous posts. No curated quotes. Verified clients only — every review is tied to a real project and a real contract.
I take on a limited number of contracts per quarter. If your project is a genuine fit, you'll receive a scoped technical proposal and fixed-price quote within 48 hours. If it's not, I'll tell you that clearly — without wasting your time.
Founders building real products with actual users or revenue targets
