Ganesh Unnam.
I build custom products for businesses, end to end: from the first requirements call to the architecture, the build, the launch and every change after it.
One person who owns the whole product, not one slice of it
I started in web development and Shopify themes, moved into React, Next.js and TypeScript on WordPress, Webflow, Sanity and Contentful, worked across IT infrastructure and cloud, and now spend most of my time on AI agent systems: orchestration, retrieval and MCP servers that connect models to the tools a team already uses.
Clients get one person from the first requirements call to the release after launch: I write the scope, design the system, build it, ship it and keep changing it as the business changes. AI is one of the things I build, not the whole story.
From the first call to the next release
One person across the whole lifecycle, so nothing gets lost between scoping, building and keeping it running.
Find the real problem before writing a line of code
I sit with the people who will use the product, map how the work happens today and agree what success looks like in numbers. Scope is cut to the smallest version that proves the idea.
- Requirements and user stories
- Scope for a first release
- Estimate and timeline
- Long-lead risks flagged early
For Sarah, the voice agent, platform approvals with long review times (Google Business Profile API, Meta app review) were flagged on day one so they would not block launch.
Design the system, the data and the cost before building it
Data model, integrations, hosting, security and running cost, written down with the trade-offs so you can see why each choice was made.
- Architecture diagram
- Data model and integrations
- Hosting and security plan
- Running-cost model
Sarah runs on Cloudflare Workers, D1 and KV with a 16-table multi-tenant schema. Modelling per-call voice costs early showed them closer to $0.50 to $1.40 a call than the first $0.05 guess, which reshaped pricing.
Ship working software every week, not a big reveal at the end
Short iterations with a demo you can click, CI/CD from the first commit and a staging site that always matches what is coming next.
- Weekly working demos
- Staging environment
- Automated tests and CI/CD
- Clean, documented code
The document pipeline was built stage by stage: OCR first, then chunking and embeddings, then retrieval and structured output, each one testable on its own.
Launch quietly, measure loudly
Production deploy, data migration, analytics and monitoring in place before the announcement, and a handover so your team knows how everything works.
- Production release
- Monitoring and analytics
- Runbook and handover
- Team walkthrough
RDPV Solutions went live as headless WordPress with a Next.js front end in Docker, DNS on Cloudflare and Google Analytics collecting only what is needed.
Keep it fast, patched and boring
Updates, security patches, uptime and performance checks on a schedule, with a short monthly note on what changed and why.
- Dependency and security updates
- Uptime and performance checks
- Backups verified
- Monthly health note
Change without breaking what already works
New requests go through one list: impact assessed, prioritised against the roadmap, built behind a release, and written up in the change log so everyone can see what moved.
- Prioritised roadmap
- Impact assessment per change
- Versioned releases
- Public change log
Sarah’s roadmap is staged: V1 review and lead modules, V2 in months 4 to 6, V3 in months 7 to 12, each release scoped against what the last one taught.
Scanned documents in, structured answers out
An OCR and retrieval pipeline on Amazon Bedrock: Textract reads the pages, pgvector finds the right passages, and the model answers in fields an app can use.
Agents, storefronts and the glue between them
What I can take off your plate
AI and voice agents
Voice agents that answer, qualify and book calls, document pipelines that read and retrieve on Amazon Bedrock, and MCP servers that connect models to your tools.
Fast websites
Next.js and React front ends in TypeScript, on WordPress, Webflow, Sanity or Contentful, that load quickly and stay easy for your team to edit.
E-commerce
Shopify and WooCommerce themes, apps and checkout tweaks.
SEO and growth
Technical SEO, structured data and the speed work that search rewards.
Automation
n8n workflows and small Node.js services that remove the copy-and-paste from your week.
Cloud and hosting
Sensible hosting, deploys and monitoring for the things above.
What I build with
AI and agents
- Claude Code
- Codex
- Cursor
- MCP
- OpenRouter
- LangChain
- RAG
- Amazon Bedrock
- Textract
Voice agents
- Vapi
- Twilio
- Call qualification
- Booking flows
- SMS
Front end
- React
- Next.js
- TypeScript
- JavaScript
- Three.js
- GSAP
- Lenis
- Vite
Back end and data
- Node.js
- PHP
- WPGraphQL
- pgvector
- Firebase
CMS and commerce
- WordPress
- Webflow
- Sanity
- Contentful
- Shopify
- WooCommerce
Cloud and delivery
- AWS
- GCP
- Cloudflare Workers
- D1
- Docker
- Kubernetes
- GitHub Actions
Automation and growth
- n8n
- Google Analytics
- Klaviyo
- SEO
Where I have done it
Gen AI Developer
RDPV Solutions · now
AI features and the website around them: LLM applications, retrieval, multi-agent workflows and MCP integrations.
Web Developer
Flooring Liquidators Canada
Storefront development and site maintenance for a multi-location retailer.
Web Developer · IT Support
AZTRA
Designed and launched aztra.ai, set up the cloud, Microsoft 365 and VoIP, and migrated email and the website with minimal downtime.
Shopify Developer
Vinyl Sound
Custom Liquid landing pages, Google Shopping, SEO and Klaviyo flows.
Web Developer
Premier Electronic Parts
Three WooCommerce stores with 15,000+ SKUs moved from cPanel to AWS, and a 25% lift in clicks and impressions in two months.
Shopify Developer
Muse Wearables, Bangalore
Theme development and storefront work for a wearables brand.
Analyst, Talent Sourcing
Deloitte
Web development and SEO
Freelance
WordPress, WooCommerce and Shopify builds and care plans for agencies and their clients.
How the work actually went
Document OCR and RAG on Amazon Bedrock
Textract OCR, chunking, embeddings and pgvector retrieval on Amazon Bedrock, returning structured answers an app can use.
Sarah, an AI employee that answers the phone
Scoping, architecting and building a voice agent that qualifies and books leads for home-services contractors, owned end to end.
Baton: one orchestrator for Claude Code, Codex and API models
Routing tasks between Claude Code, Codex and API models with shared context, through plan, validate, code, test and integrate stages.
talentsear.ch, x-ray search for recruiters
X-ray search for recruiters and job seekers, with predefined filters across professional and creative platforms.
RDPV Solutions, headless WordPress on Next.js
Headless WordPress with a Next.js front end in Docker, behind Cloudflare DNS, for a technology, design and talent company.
Three ways to start
Discovery sprint
Requirements, architecture and a costed plan, so you know exactly what you are buying before the build starts.
- Stakeholder sessions
- Requirements and scope
- Architecture and cost model
- Fixed estimate for the build
Build and launch
Design, development and launch in short iterations with a demo every week.
- Weekly demos on staging
- CI/CD and tests
- Launch, monitoring, handover
Keep and evolve
Maintenance, monitoring and a steady flow of changes against a shared roadmap.
- Updates and security patches
- Change requests with impact notes
- Roadmap and change log