All work

Confidential AI product engineering

AI-Powered Sales and Operations Platform

I built substantial parts of a private AI-assisted sales and operations platform spanning CRM workflows, quotations, tasks, messaging, payments, referrals, dashboards, localization, and production AI/chat integrations.

Role
Full-Stack Product Engineer through HestiaForge
Period
2025–2026

Confidential product system

AI assistance inside operational workflows

Simplified system map

Client and CRM workflows

Quotations and documents

AI and streaming chat

Payments and billing

Evidence-led workflow diagram; not a recreated private interface or architecture disclosure. Diagram connecting CRM records, quotations, AI chat, payments, and operations into a production sales and operations product flow.

Context

Private client product

Role

Full-Stack Product Engineer through HestiaForge

Product type

Confidential AI product engineering

Core capability

React

Product context

Keeping a fast-moving operational product coherent

The product brings together client records, sales workflows, calculation and document processes, communications, billing, learning content, and operational visibility. The challenge was to connect conventional workflows with useful AI assistance without turning the product into disconnected demos.

Attribution anchor

What I was responsible for

I worked across the existing team product and its generated foundations, taking ownership of substantial feature areas and production fixes. I do not present the platform as a solo build, and its identity and private workflow details remain confidential.

Selected scope

Selected contributions

  1. 01

    Implemented quotation and calculator workflows, client lookup and deduplication, tasks, alerts, referrals, and presentation templates.

  2. 02

    Integrated specialist product and document workflows, including proposal and policy extraction, while keeping the public description intentionally generic.

  3. 03

    Built messaging flows, streaming chat, conversation history and memory, and application integration with AI and agent tooling.

  4. 04

    Delivered subscription, credit, public checkout, payment, and advisory-billing workflows.

  5. 05

    Improved dashboards, localization, training and analytics surfaces, authentication, and production reliability.

Evidence, not reconstruction

System view

A simplified view of verified responsibilities and system relationships. It does not reproduce a private or historical interface.

Confidential product system

AI assistance inside operational workflows

A generic, evidence-led view that intentionally omits the private product interface and domain specifics.

Simplified system map

Client and CRM workflows

Records, lookup, deduplication, tasks, alerts, and referrals.

Input

Quotations and documents

Calculations, presentations, proposal and document extraction.

Process

AI and streaming chat

Conversation history, memory, agent integration, and reviewable assistance.

Process

Payments and billing

Subscriptions, credits, checkout, and advisory billing workflows.

System

Operations and analytics

Dashboards, localization, training, analytics, and reliability.

System

Production product flow

Connected capabilities usable inside day-to-day sales and operations.

Output

Verified relationships

  • Client and CRM workflowsQuotations and documents
  • Quotations and documentsAI and streaming chat
  • Client and CRM workflowsPayments and billing
  • AI and streaming chatOperations and analytics
  • Payments and billingProduction product flow
  • Operations and analyticsProduction product flow
Evidence-led workflow diagram; not a recreated private interface or architecture disclosure. Diagram connecting CRM records, quotations, AI chat, payments, and operations into a production sales and operations product flow.

Engineering judgment

Selected technical decisions

Decision 01

Used Supabase/PostgreSQL and Edge Functions as a practical backend for authentication, data, and workflow services.

Decision 02

Connected AI SDK and agent-based chat capabilities to existing operational workflows rather than presenting AI as a separate demo.

Decision 03

Combined typed forms and validation, React Query state, analytics, and incremental production hardening to support rapid product iteration.

Technology context

  • React
  • Vite
  • TypeScript
  • Supabase
  • PostgreSQL
  • Edge Functions
  • React Query
  • AI SDK
  • Mastra
  • Payment workflows
  • PostHog
  • D3 / Recharts
  • React Hook Form
  • Zod
  • i18n

Delivered capability

Connected product areas moved from iteration into production-ready operation across frontend, data, workflows, payments, integrations, localization, analytics, and reliability.

AI product features · Operational workflows · Full-stack delivery

Need AI features embedded into a serious operational product?