portfolio
Craig Bosman
Full-Stack Developer & AI Systems Specialist
📍 EST Timezone | 💼 Available for Contract Work | 🚀 Building Production AI Systems ⭐ Upwork Rising Talent (Top 5% of New Freelancers - Feb 2026)
👋 About Me
I build production AI systems — full-stack products, automated pipelines, and intelligent infrastructure that runs without babysitting. My work spans consumer products (a live chess coaching app for kids), B2B automation (a fully hands-off intelligence platform for financial services clients), and large-scale data systems (71 microservices, 54K events/day, paying subscribers).
What makes me different:
- 🏗️ Production standards — I ship systems with 99%+ uptime, real transactions, and comprehensive error handling. Not MVPs. Not demos.
- 🤖 AI architecture depth — I design AI pipelines where models do the right job: deterministic engines for analysis, LLMs for language. Hallucination-proof by design.
- 🎯 Full-stack ownership — From Next.js frontend to Python backend to Railway/AWS deployment. One person, complete delivery.
- 💡 Teaching background — I homeschool five kids and built an edtech product from that experience. I understand how people learn, not just how systems work.
🚀 Featured Projects
Pawnprint - AI Chess Coaching Platform
Live at pawnprint.com | 99% uptime | 5.4M puzzles | 8,543-chunk RAG corpus
AI chess coaching platform for kids where LLM hallucination is prevented architecturally — Stockfish evaluates positions, a deterministic rule engine maps output to teaching concepts, Gemini Flash translates to age-appropriate language. Never the other way around.
- Scale: 5.4M indexed Lichess puzzles, 8,543 enriched RAG chunks from classical chess literature
- Reliability: 99% uptime, $60/month infrastructure, 100% Phase-0 tester retention
- Innovation: Three-layer deterministic + LLM pipeline; per-user vocabulary acquisition tracking across 7 exposures
- Tech: Next.js 14, TypeScript, Turso, Stockfish 18, Gemini 2.5 Flash, SM-2, Glicko-1
| View Details | COPPA Case Study → |
TalentSignals - B2B Competitive Intelligence Platform
4/4 milestones delivered | 39/39 tests passing | Fully automated monthly pipeline
Fully automated pipeline delivering monthly branded PDF competitive intelligence reports to GCC financial services clients. Four-stage architecture: Enrichlayer + Apify + Playwright data collection → two-pass Claude API analysis → ReportLab PDF generation → SendGrid delivery with Stripe billing.
- Automation: Fully automated monthly cycle after client onboarding
- AI Pipeline: Two-pass Claude architecture — structured extraction then insight generation
- Reliability: 39/39 tests passing, live Stripe billing, delivery audited via SendGrid webhooks
- Tech: Python, FastAPI, Supabase, Railway, Claude API, ReportLab, Stripe, SendGrid
Anubis Bot - Cryptocurrency Intelligence Platform
99% uptime | $480/month MRR | 54,000 events/day | 🗄️ Archived
71-microservice production system that processed cryptocurrency market data in real-time, serving 31 paying subscribers via Stripe integration.
- Scale: 29GB PostgreSQL database, 1.86M rows, 54K events/day
- Reliability: 99% uptime over 220 consecutive days, zero data loss
- Business: $480/month recurring revenue, real transactions, Stripe payments
- Tech: Python (asyncio), PostgreSQL, Telegram bots, Docker, SystemD
| View Details → | Read Case Study → |
FreelanceOS - AI-Powered Upwork Automation
80% time reduction | 10-minute response time | Mobile-first workflow
End-to-end automation system reducing Upwork proposal work from 10+ hours/week to under 2 hours. Built with n8n, Claude Sonnet 4.5 API, ClickUp, and Telegram Bot.
- Impact: 8-10 hours/week saved ($800-1,000/week value)
- Tech: n8n, Claude API, ClickUp API, Telegram Bot, Docker, PostgreSQL
- Innovation: External prompts system - edit AI behavior via text files, no code changes
| View Details → | Read Case Study → |
Docira - AWS Serverless RAG System
90-minute build | $24/month infrastructure | 100+ MB PDFs
Complete AWS serverless RAG pipeline built in 90 minutes despite never using AWS before. Production-ready on first deployment.
- Speed: Concept to production in 90 minutes (learning AWS from scratch)
- Scale: Handles 100+ MB PDF files, multi-level extraction fallback
- Cost: $24/month AWS costs (Lambda + S3 + API Gateway)
- Tech: AWS Lambda, S3, API Gateway, Claude API, React, Tailwind CSS
Praxis Bot - Solana DeFi Monitoring
99.9% accuracy | Zero data loss | 30-second real-time updates | 🗄️ Archived
Production Solana blockchain monitoring system with 99.9% accuracy versus professional trading platforms.
- Accuracy: Reversed-engineered 281-byte binary blockchain data format
- Reliability: 99.9% uptime across 15+ deployments, zero data loss
- Scale: 50+ positions per wallet, 30-second update intervals
- Tech: Solana WebSocket API, Binary parsing, Telegram integration
| View Details → | Read Case Study → |
Financial Data Pipeline - Multi-API Integration
OAuth 2.0 | 1,000+ records | Client delivery
Multi-API orchestration system integrating financial data APIs with automated Google Drive exports. Delivered for paying client.
- Integration: OAuth 2.0 authentication, Google Drive API, financial data APIs
- Web Interface: Flask dashboard, role-based access control
- Status: Delivered for paying client ($1,250 contract)
- Tech: Python, Flask, OAuth 2.0, Google Drive API, PostgreSQL
🛠️ Development Methodology
Praxis SDD - Spec-Driven Development for AI-Assisted Builds
Open source | MIT licensed | In active production use
A documented methodology for building software with AI coding agents without shipping wrong things. Combines a pre-implementation articulation layer (constitution → spec → plan → tasks) with investigation-first phase execution, approval gates, and cross-artifact consistency verification at closure.
- Origin: Derived from 21-phase COPPA compliance work for Pawnprint — discipline caught a Phase 3.5 spec-drift error before silent closure
- Includes: 17 files covering the full build lifecycle, 3 real filled examples, token efficiency rules for Claude Code
- Compatible: Claude Code, Cursor, Copilot Workspace, Codex CLI, Gemini CLI
-
Tech: Markdown templates Claude Code Any AI coding agent
💼 What I Do
Core Competencies:
- 🤖 AI Integration - Claude API, Gemini, multi-model architectures, RAG pipelines
- ⚡ Full-Stack TypeScript - Next.js 14 App Router, Turso/libSQL, production deployments
- 🔄 Workflow Automation - n8n, event-driven architectures, zero-touch pipelines
- 📊 API Development - RESTful APIs, webhooks, OAuth 2.0, multi-service orchestration
- ☁️ Cloud Infrastructure - AWS Lambda/S3/API Gateway, Railway, DigitalOcean, Supabase
- 🗄️ Database Design - PostgreSQL optimization, Turso/libSQL, large-scale data management
- 🔗 Blockchain/DeFi - Solana, binary data parsing, real-time on-chain monitoring
- 🔒 Privacy/Compliance - End-to-end COPPA compliance under the amended 2025 Rule, processor data processing agreements, written information security programs, append-only audit infrastructure, plan-ladder posture scaling — engineering discipline transfers to HIPAA, FERPA, GDPR, GLBA, PCI-DSS
Languages & Tools:
- Languages: Python (expert), TypeScript, JavaScript/Node.js, SQL
- Frameworks: Next.js, React, Flask, FastAPI, n8n, Docker
- AI/ML: Claude API, Gemini API, OpenAI, RAG pipelines, Stockfish integration
- APIs: Stripe, SendGrid, Telegram, ClickUp, Enrichlayer, Apify
- Databases: PostgreSQL, Turso (libSQL), Supabase, Redis
- DevOps: Railway, DigitalOcean, AWS, Docker Compose, PM2, Cloudflare Tunnel
📊 By The Numbers
| Metric | Achievement |
|---|---|
| Upwork Recognition | ⭐ Rising Talent (Top 5% of New Freelancers - Feb 2026) |
| Production Uptime | 99%+ (Pawnprint active, Anubis Bot 220 days, Praxis Bot 15+ deployments) |
| Live Product | pawnprint.com — 5.4M puzzles, 8,543 RAG chunks, 100% tester retention |
| Client Delivery | TalentSignals — 4/4 milestones, 39/39 tests, fully automated pipeline |
| Revenue Generated | $480/month MRR (Anubis Bot subscribers) |
| Time Saved | 8-10 hours/week (FreelanceOS automation) |
| Data Scale | 54,000 events/day, 29GB database, 1.86M rows |
| Accuracy | 99.9% (Praxis Bot vs professional platforms) |
| Cost Efficiency | $24/month (Docira serverless), $60/month (Pawnprint full-stack) |
| Children’s Privacy | End-to-end COPPA compliance under amended 2025 Rule — 21 production phases, 8 processor DPAs, formal § 312.4–312.10 self-assessment, append-only audit, end-to-end verified in production |
🎯 How I Work
My Process:
- Understand the business problem - Not just technical requirements
- Design for production - 99%+ uptime, real transactions, zero data loss
- Build iteratively - Ship fast, iterate based on real usage
- Optimize for cost - Lean infrastructure that scales without surprise bills
- Own the full stack - Frontend to backend to deployment, no handoffs
What You Get:
- ✅ Production-grade code (not MVPs or demos)
- ✅ Comprehensive error handling (graceful degradation)
- ✅ Real-world testing (edge cases, concurrent access)
- ✅ Cost-optimized architecture (no unnecessary spending)
- ✅ Same-day responses (EST timezone, <1 hour during business hours)
📫 Get In Touch
Availability: Open to contract work (fixed-price or hourly)
Start Here:
- 💡 Book $75 Technical Consultation - 30-minute feasibility assessment (credited toward project if hired)
- 💼 Hire on Upwork - Rising Talent freelancer
- 📧 Email: cbosman@praxisprotocol.io - Direct contact
Best For:
- AI product development (RAG pipelines, multi-model architectures, edtech)
- Full-stack TypeScript applications (Next.js, production deployments)
- Automated B2B pipelines (data collection → analysis → delivery)
- Production systems requiring 99%+ uptime
- AWS serverless architectures (Lambda, cost-optimized)
- API integration (multi-service orchestration, OAuth 2.0)
- Workflow automation (n8n, Zapier, complex orchestration)
- COPPA / children’s privacy compliance (plan ladder articulation, processor DPAs, formal self-assessments, audit infrastructure)
- Regulated-industry compliance posture (HIPAA, FERPA, GDPR, GLBA — same disciplined execution patterns, different subject matter)
Response Time: <1 hour during business hours (9 AM - 6 PM EST)
Other Platforms:
- 🔗 LinkedIn: Craig Bosman
- 💻 GitHub: @bozzdev
🏆 Why Work With Me
Production Standards: Most developers ship working code. I ship systems with 99%+ uptime, real error handling, and zero data loss — proven across multiple production deployments.
AI Architecture Depth: I design AI pipelines correctly. Pawnprint’s coaching system prevents hallucination by construction — Stockfish handles analysis, the LLM handles language. That separation matters in systems where accuracy counts.
Full-Stack Ownership: Next.js frontend, Python/FastAPI backend, Railway/AWS/DigitalOcean deployment. One person handles the full stack — no coordination overhead, no handoffs.
Business Thinking: I understand ROI, not just code. Built $480/month MRR, delivered fully automated B2B pipelines, designed systems where variable cost per client is predictable.
Communication: EST timezone with <1 hour response times. Same-day responses, daily updates, proactive problem notification.
Proven Track Record:
- 99% uptime, 100% tester retention (Pawnprint - live at pawnprint.com)
- 4/4 milestones delivered, 39/39 tests passing (TalentSignals - complete)
- 99% uptime over 220 days (Anubis Bot - archived)
- 99.9% accuracy vs professional platforms (Praxis Bot - archived)
- 80% time reduction (FreelanceOS - active)
- $24/month serverless architecture (Docira - active)
📚 Case Studies
Pawnprint COPPA Compliance: A Children’s Privacy Implementation Under the Amended 2025 Rule
Comprehensive end-to-end COPPA compliance build for a children’s-data-handling commercial service — application infrastructure, procedural artifacts, and the engineering discipline that made both survive audit
Twenty-one production phases across two structured workplans, built specifically against the amended 2025 Rule under the compliance deadline that just passed. Application infrastructure shipped: verifiable parental consent, six parental rights surfaces, two-tier authentication, append-only audit infrastructure enforced at the application code level, soft-delete-then-purge with anonymization on purge. Procedural infrastructure shipped: eight processor data processing agreements, privacy notice statically rendered from committed source, written data retention policy, formal § 312.4 through § 312.10 self-assessment, operative posture closure document, comprehensive self-audit. End-to-end-verified in production with cross-document consistency validated at closure (15-of-15 cross-artifact + 10-of-10 operational probes passing). The engineering discipline transfers to HIPAA, FERPA, GDPR, GLBA, PCI-DSS, and any regulated-industry compliance build.
Key Highlights:
- 21 production phases across Workplan 2 (13 phases, application infrastructure) and Workplan 3 (8 phases, procedural artifacts)
- Built against the amended 2025 Rule, not the legacy 2013 baseline
- Eight processor data processing agreements with documented written assurances per § 312.8(c)
- Formal § 312.4–312.10 self-assessment (~628 lines) plus operative posture closure document plus comprehensive self-audit (~828 lines)
- End-to-end verified in production against real closed-beta cohort with full append-only audit trail across multiple policy versions
- Plan ladder articulated with explicit escalation triggers — escalation paths from current posture to attorney engagement, expanded engagement, or Safe Harbor program participation, each with documented activation conditions
- Engineering discipline transfers to HIPAA, FERPA, GDPR, GLBA, PCI-DSS, and any regulated-industry compliance build
FreelanceOS: 80% Time Reduction Through AI Automation
How I automated my entire Upwork workflow from 10+ hours/week to under 2 hours using n8n, Claude API, and Telegram Bot
Complete deep dive into the technical architecture, implementation timeline, and business impact of building an intelligent job screening and proposal generation system.
Key Highlights:
- 100-point job evaluation matrix
- Automated proposal generation with 95%+ first-draft quality
- Real-time Telegram notifications
- Time savings breakdown and ROI analysis
Anubis Bot: Building a $480/month SaaS in Production
Lessons learned building and maintaining a 71-microservice cryptocurrency platform serving 31 paying subscribers
Complete story of building a production-grade cryptocurrency intelligence platform that achieved 99%+ uptime over 220 days while processing 54,000 events daily.
Key Highlights:
- 71 microservices orchestrated via SystemD
- 29GB PostgreSQL database with 1.86M tokens tracked
- 5-model ML ensemble with outcome tracking
- Stripe + Solana Pay revenue implementation
- Strategic pivot analysis and business lessons
Praxis Bot: 99.9% Accuracy Through Binary Data Reverse Engineering
How I reverse-engineered Solana’s 281-byte binary format and achieved 99.9% accuracy vs professional platforms
Complete documentation of reverse-engineering methodology for parsing Solana blockchain transaction binaries.
Key Highlights:
- 281-byte binary format completely reverse-engineered
- 99.9% accuracy validated against 1.5M+ transactions
- Sub-millisecond parsing latency
- Competitive comparison vs Helius, QuickNode, dexscreener
💡 Technical Feasibility Consultations
Not sure if your AI/automation project is technically feasible? Get expert assessment before committing to full development.
30-Minute Technical Assessment - $75
What You’ll Get:
- ✅ Technical approach recommendation (specific technologies & architecture)
- ✅ Timeline & cost estimates (realistic, not theoretical)
- ✅ Risk assessment (what could go wrong, how to mitigate)
- ✅ Go/no-go recommendation (honest evaluation)
- ✅ Written summary with actionable next steps
💰 Consultation Fee Credit: If you hire me for the full project within 30 days, your $75 consultation fee is fully credited toward the project cost. Zero risk.
Process:
- Submit project details via Upwork
- I review and prepare assessment (24 hours)
- 30-minute consultation call (Zoom/Google Meet)
- Receive written assessment within 2 business days
Best For:
- 🏢 Teams evaluating AI integration feasibility (Claude API, RAG pipelines, automation)
- 💼 Companies needing expert validation before budget approval
- 🔍 Projects requiring independent technical assessment
- 🤔 Second opinions on proposed technical approaches
- ⚡ Fast evaluation needed (assessment within 48 hours)
Why Trust My Assessment?
Production-grade expertise, not theoretical knowledge:
- Built and shipped a live AI product (pawnprint.com) with 99% uptime
- Delivered fully automated B2B pipeline across four milestones for paying client
- Built and operated 71-microservice cryptocurrency platform over 220 days
- Real experience with AI integration (Claude, Gemini, multi-model architectures)
I give honest recommendations - if your project isn’t technically feasible or cost-effective, I’ll tell you upfront.
📅 Book Your Technical Consultation
Or email me for more info: cbosman@praxisprotocol.io
🎓 Background
Experience:
- Full-stack development (TypeScript/Python, 6+ years)
- Production AI systems architecture
- AI/ML integration (Claude API, Gemini, multi-model pipelines)
- Teaching background — homeschool educator, 5 kids, built edtech product from lived experience
Domain Expertise:
- EdTech (Pawnprint — chess coaching, age-adaptive AI, spaced repetition)
- Finance/Trading (market data, real-time processing, DeFi)
- B2B Intelligence (competitive analysis pipelines, automated reporting)
- Healthcare/Insurance (HIPAA compliance understanding)
- Blockchain/Cryptocurrency (Solana, DeFi, on-chain analysis)
Personal:
- Homeschool parent (5 kids)
- EST timezone (North Carolina)
- Focused development blocks — async-friendly, reliable communication
Let’s build something together. 🚀
| Book $75 Consultation | Contact Me | View Projects |