Real systems, clear technical execution

Selected Product Builds & Case Studies

Security audits that caught real risks before launch, production systems I've shipped, and platform work for agencies — each one showing how I approach architecture, integrations, automation and production-ready execution.

Technical Audits & Security Reviews

Independent reviews of existing production systems — finding the security, correctness and cost risks that matter before a product scales.

AI Platform Technical Audit

AI Agent Platform for Gyms

A technical review of an AI agent platform already running in production across multiple gyms, focused on surfacing security, reliability and cost risks before the system was scaled further.

Found duplicated LLM calls inflating both cost and latency
Identified secret keys exposed in the codebase
Flagged sensitive routes shipping without proper protection
Delivered prioritized, practical fixes ahead of scaling

~50%

lower AI costs

reliability bugs confirmed

unprotected routes flagged

hardcoded secrets caught

What I Handled

Reviewed existing AI workflows and implementation
Identified duplicated LLM calls increasing cost and latency
Identified exposed secret keys
Identified sensitive routes without proper protection
Reviewed reliability and maintainability risks
Provided practical recommendations for improvements

Outcome

Surfaced concrete production risks around security, cost and route protection, and clarified exactly what needed to be fixed before the platform could be safely scaled.

Technical Focus

AI agent workflows, LLM usage, security review, route protection, API safety, reliability, maintainability, cost control.

Stack / Technologies

AI agentsLLM APIsBackend routesAPI integrationsSecurity reviewWorkflow analysis

Full-Stack Security Audit

Booking & User-Management Platform

A full-stack security audit of a production booking and user-management platform (Next.js + Supabase, with a WhatsApp bot). Every finding was documented with file-level evidence, a severity rating and a verification test.

Found a critical data-exposure flaw: a public endpoint leaked sensitive user records to anyone guessing an email, phone or ID
Flagged unauthenticated uploads that could write files into any user's record
Caught fail-open cron and webhook auth that let unauthenticated callers trigger backups, campaigns and booking changes
Mapped inconsistent identity matching that risked updating the wrong user's record
Delivered severity-ranked findings with file-level evidence, fixes and a verification test each

user-record leak caught

critical vulnerability

high-severity gaps

reliability risks flagged

What I Handled

Full-stack security review (Next.js app + Supabase backend + WhatsApp bot)
Data-exposure review of public endpoints and service-role usage
API authorization review (RLS vs service-role consistency)
Webhook and cron hardening (fail-closed design)
Architecture and code-duplication review
Performance and reliability review

Outcome

Gave the team a prioritized, test-backed remediation plan focused first on protecting sensitive user data and closing unauthenticated write paths before scaling.

Technical Focus

Sensitive-data protection, public-endpoint authentication (signed tokens), authorization consistency, webhook signature verification, fail-closed design, identity resolution, reliability.

Stack / Technologies

Next.jsSupabasePostgreSQL + RLSWhatsApp APIService-role authCron jobs

Web3 Security & Architecture Audit

Web3 Energy-Tokenization Platform

A full security and architecture review of a production Web3 staking platform — smart contracts, dApp and on-chain indexer. Every finding was documented with code-level evidence, a severity rating and a concrete remediation path.

Uncovered a critical custody flaw — user wallets were fully controlled by a backend admin key, not the users
Identified a payment-integrity exploit: buy-then-refund while keeping minted tokens (no idempotency, no revocation)
Caught smart-contract accounting drift from floor division and mutable conversion rates on active stakes
Flagged server actions trusting client-supplied identity (userId / email / wallet) with no session binding
Surfaced an exposed admin-key path in client-side code, an open indexer API and missing security headers
Delivered prioritized findings (Critical / High / Medium) plus three architectural remediation options

fund-drain risk caught

15+

vulnerabilities found

41%

gas optimization

88%

indexing cost reduction

What I Handled

Full-stack security & architecture review (Solidity + Next.js dApp + indexer)
Smart contract audit: custody, staking math, ownership and multisig
Payment lifecycle review: Stripe webhooks, idempotency and refunds
Server-side authorization review and identity binding
Infrastructure review: secrets handling, security headers, blob storage, indexer API
Prioritized remediation roadmap with architectural options

Outcome

Gave the team a clear, evidence-backed picture of where user funds and payments were at risk — and three concrete paths forward (harden the custodial model, move to user-controlled smart accounts, or a hybrid) so they could decide how to proceed before scaling.

Technical Focus

Smart contract security, custodial vs non-custodial architecture, account abstraction (ERC-4337), payment idempotency, server-side authorization, on-chain indexing and observability.

Stack / Technologies

SolidityviemNext.jsAuth0StripeInngestPonder (indexer)PostgreSQL

Product & Platform Builds

Products and platforms I designed and built — from revenue-generating SaaS to reusable infrastructure operated in real environments.

Production SaaS — designed, built & launched

CosloAI

Live

A multi-channel AI SaaS platform I designed, built and launched from scratch, letting businesses deploy custom AI assistants across websites, WhatsApp, Instagram, Facebook and Shopify.

Live SaaS with real users and paying customers
Assistants running across 5+ channels from one platform
Billing, usage metering and admin tooling in production
Owned end-to-end: architecture → deployment → operations

20+

paying users

1000+

conversations handled

92%+

conversation success

417+

bookings confirmed

What I Handled

Product architecture
Multi-tenant SaaS structure
AI assistant workflows
RAG / knowledge base logic
Channel integrations
Billing and subscription logic
Usage tracking
Dashboards and internal management tools
Deployment and production infrastructure

Outcome

Taken from idea to a deployed, revenue-generating product — including architecture, channel integrations, billing, admin tooling and the operational workflows that keep it running.

Technical Focus

AI workflows, RAG, multi-channel messaging, SaaS architecture, billing, dashboards, production infrastructure.

Stack / Technologies

Node.jsTypeScriptPostgreSQLReactAI APIsRAG workflowsMessaging integrationsPayment/subscription logic

Multi-Tenant AI Agent Platform — for an agency

MetaGateway

A platform I'm designing and building for an agency to replace its old model of hand-building a new codebase for every client. One centralized Meta gateway (Lead Ads + WhatsApp), a reusable per-client CRM, a typed CRM–agent contract and a safe agent-execution model — configured per client instead of rebuilt each time.

Replaced one-off per-client codebases with a single configuration-driven platform
Centralized multi-tenant Meta gateway so client CRMs never touch Meta directly
Typed CRM–agent contract with a safe execution model: the CRM validates every AI action, agents make no autonomous writes
Deterministic-first design — state machines for flows, LLM only for guarded FAQ with confidence thresholds and human handoff
Config-driven onboarding (observe → draft → auto) with runbooks and smoke tests so junior devs can operate it

What I Handled

Platform and multi-tenant architecture
Agency-owned Meta gateway (OAuth, webhooks, Lead Ads + WhatsApp Cloud API)
Reusable per-client CRM template (backend + frontend)
Versioned, typed CRM–agent contract
AI agent execution and safety / guardrail model
Security model: HMAC-signed boundaries, encrypted token storage, scoped keys
Runbooks, smoke tests and deployment documentation

Outcome

Moves the agency from bespoke builds to repeatable delivery — faster client onboarding, fewer codebases to maintain, safer AI automation and clear boundaries for junior developers. Currently in pilot rollout with the first live client agent.

Technical Focus

Multi-tenant platform architecture, Meta Lead Ads / WhatsApp Cloud API integration, HMAC-signed service boundaries, deterministic-vs-LLM design, AI action validation and guardrails, configuration-driven delivery.

Stack / Technologies

TypeScriptNode.jsPrismaZodVitestWhatsApp Cloud APIClaudeHMAC webhooks

Client Web3 Product — built to production

DecaWin

Live

A client Web3 product I built from scratch and took to production — frontend, backend, wallet flows and the on-chain integrations needed to support the launch.

Shipped a full Web3 product to production from zero
Live wallet connection, presale and staking flows
On-chain / off-chain event indexing and synchronization
Admin dashboard and backend services for launch ops

What I Handled

Full-stack implementation
Wallet connection flows
Smart contract interaction
Token/presale-related functionality
Admin dashboard functionality
Backend services
Event indexing and on-chain/off-chain synchronization
Deployment and production setup

Outcome

Delivered the full technical product from scratch and brought it to production, from wallet flows and smart-contract interaction to backend services and deployment.

Technical Focus

Web3 product delivery, wallet integrations, smart contract interaction, backend systems, production deployment.

Stack / Technologies

SolidityTypeScriptReactNode.jsWeb3 librariesWallet integrationsBackend services

In-House R&D Builds

Complete systems I designed and built end-to-end to develop and pressure-test specific architectures — built to the same standards as client work.

In-House R&D Build

AgencyOps

Live

A full agency-operations system covering how inbound leads are captured, qualified and managed with AI-assisted workflows, structured data, webhook automation and operational dashboards.

End-to-end pipeline: lead intake → AI qualification → dashboard
AI summaries and suggested replies wired into the workflow
Webhook automation, notifications and audit logging

What I Handled

Product architecture
Lead intake flow
AI-assisted qualification workflow
Dashboard UI
Backend logic
API and webhook integrations
Audit/logging and internal workflow logic

Outcome

A working reference implementation for agency-facing internal tools — showing how I structure automation workflows and dashboard-based products from architecture through implementation.

Technical Focus

Agency operations, lead qualification, AI-assisted workflows, dashboards, backend automation, API/webhook integrations.

Stack / Technologies

Node.jsTypeScriptReactPostgreSQLAPI integrationsWebhook automationAI workflows

In-House R&D Build

VoiceAI

Live

A real-time voice intake system that answers inbound calls, captures structured lead details through a conversational flow, and creates follow-up automation.

Real-time phone intake with structured data capture
Conversational AI flow wired to storage and follow-up
Built on a production voice stack (Twilio + OpenAI Realtime + Deepgram)

What I Handled

Workflow design
AI interaction logic
Backend implementation
API integrations
Structured data capture
Follow-up automation flow

Outcome

A working system connecting live call input, AI logic, data storage and follow-up actions into one maintainable product — the kind of real-time AI workflow I build for clients.

Technical Focus

AI intake workflows, conversational automation, backend logic, structured data capture, integrations, follow-up automation.

Stack / Technologies

TwilioOpenAI RealtimeDeepgramNode.jsTypeScriptReact

Client Feedback

Trusted by founders and teams who came back

Across 30+ projects, the same themes come up: clear communication, strong technical judgment, and someone who stays involved after delivery — not just another order taker.

A pleasure to work with. Great communication and deep involvement. He proposed features we didn't even ask for and took the time to explain every option in detail whenever there was a decision to make. We'll work together again, for sure.

Client · France

A talented, dedicated person you can build a trusted collaboration with. Communication is great and responsive, and getting the project done was stressless, simple and pleasant. Most important to me: he stays available after delivery to help whenever needed.

Client · France

One of the best experiences I've had. He helped with everything along the way and fully thought through the implementation, asking the right questions to make sure his delivery met my needs. I wouldn't consider anyone else if you value timeliness, knowledge and someone who knows what they're doing.

Client · United States

A great experience. He showed a clear understanding of our requirements and brought a fresh perspective that significantly improved the direction of our solution. He added features that enhanced the app and went above and beyond our expectations.

Client · United Kingdom

Outstanding service. Very professional, high-quality work, delivered precisely to the requirements — and extremely fast. Would definitely work with him again.

Client · United States

Not the first time I've worked with him, and definitely not the last. He asks a lot of the right questions and delivers exactly what's needed — every single time. Can recommend 100%.

Client · Czech Republic

Next Step

Need a technical partner for something similar?

I help founders, agencies and small teams build, fix and maintain SaaS products, AI workflows, internal tools, dashboards, integrations and Web3 products without hiring a full-time developer.

Discuss your Project View services