AI-native SaaS · 2026
Waaru
A WhatsApp automation platform that keeps conversations on-brand, resumes interrupted flows and brings humans in at the right moment.
From agent state machine to production SaaSBengaluru, India · Building AI that works beyond the demo
Tanay Patel.
Full-stack & Generative AI Engineer
I design and ship AI products that hold up in production, from RAG and agent workflows to the full-stack software around them.
Software Engineer 2 · AI Architect
- 6+
- years building production software
- AI × Full-stack
- architecture through delivery
- Build · secure · ship
- one connected engineering loop
Engineering approach
01 / APPROACH
A good AI demo is the starting line. The real work is making it reliable, secure and useful every day.
I work across evaluation, security, product UX, APIs and cloud delivery to turn a promising model response into a product people can rely on.
02 / SELECTED WORK
Systems built for the
messy real world.
A few of the products I have helped shape, including AI support tools, retail software, invoicing systems and e-commerce platforms.
→
Enterprise AI · Zendesk Labs
AI support learning systems
Goal-based training experiences that turn historical support context into adjustable, AI-generated customer scenarios and behaviours.
Generative AI applied to repeatable support-agent developmentRetail SaaS · 2025
POS billing & inventory platform
A point-of-sale system for shops to manage invoices, payments, customer credit, inventory and reporting from one operational workspace.
Connected accounting, WhatsApp and IVR workflows for daily retail operationsFintech SaaS · 2023
InvoiceIndia
An invoicing platform for Indian businesses to generate GST e-invoices and e-way bills through integrated tax-department APIs.
Connected compliance, accounting and WhatsApp Business communicationE-commerce · Multi-brand retail
Multi-brand e-commerce platform
A redesigned commerce system with multiple storefronts managed through one unified administration console.
Integrated payments, logistics, CMS, WhatsApp, technical SEO and AWS delivery03 / EXPERTISE
Across the stack.
Better together.
I connect the model, interface, backend and infrastructure so the product behaves like one system rather than four separate ones.
Production AI systems
RAG, evaluation pipelines and secure LLM integrations that still work when the inputs get messy.
- RAG
- LLM evaluation
- AI security
Agentic workflows
Stateful agents with tool use, human handoffs and a clear plan for failure and recovery.
- Agents
- HITL
- AWS Bedrock
Full-stack products
Product discovery, useful interfaces, resilient APIs and cloud delivery treated as one system.
- Next.js
- Node.js
- AWS
04 / TOOLKIT & CREDENTIALS
Depth across the stack.
Judgment across the system.
Tools change quickly. The useful constant is knowing where to apply them, what can fail and how to make the whole product hold together.
AI engineering
Building AI systems that teams can evaluate, secure and operate after launch.
- RAG
- Agentic workflows
- LangChain
- LLM evaluation
- Vector databases
- Prompt engineering
- AI security
- AWS Bedrock
Product frontend
Making complex workflows feel clear, fast and predictable for the people using them.
- TypeScript
- JavaScript
- React
- Next.js
- PWA
- Browser extensions
- Design systems
- Analytics
Backend & cloud
Designing the APIs, data models and cloud infrastructure that keep products dependable.
- Node.js
- Express
- Python
- AWS
- Kubernetes
- MongoDB
- MySQL
- DynamoDB
CERTIFICATIONS
Formal foundations,
applied in practice.
Microsoft
Programming with JavaScript
NPTEL · IIT
Information Security IV
NPTEL · IIT
Information Security V
Coursera
Search Engine Optimization
05 / EXPERIENCE
A career shaped by
increasing ownership.
From building complete products in small teams to creating and governing enterprise AI capabilities.
2024—Present
Zendesk
Software Engineer 2 · AI ArchitectBuilding and assessing AI-powered products in Zendesk Labs—from generative training systems and ticket intelligence to LLM comparison tooling.
- Enterprise AI
- Product architecture
- AI security
2021—2024
Snaphunt
Software EngineerShipped full-stack recruiting workflows, intelligent hiring automation and multi-channel communications inside a fast-moving remote startup.
- SaaS
- Full-stack delivery
- Product analytics
2019—2021
Independent & early teams
Full-stack DeveloperBuilt e-commerce and conference platforms, connected payments, logistics, CMS and messaging systems, and deployed them on AWS.
- E-commerce
- Integrations
- Cloud delivery
06 / BEYOND THE BUILD
Life beyond the
keyboard.
I enjoy travel, unfamiliar places and long conversations that have nothing to do with software. Time away from the screen usually helps me return with a clearer sense of what matters.
Based in BengaluruOften travellingAlways curious
07 / FIELD NOTES
Notes from
real builds.
What broke, what changed my mind and the patterns I would use again while building production AI and full-stack products.
View all writingProduction AI begins after the demo
What I look for when an AI feature has to survive real users, messy inputs and broken dependencies—not just a polished demo.
Jul 2026Building Production AI · 1
Stateful agents for real conversations
What building conversational agents taught me about explicit state, recovery paths and human handoffs outside a controlled demo.
Jul 2026Building Production AI · 2
Design SaaS boundaries before features
Five boundaries I define early so a full-stack SaaS product can grow without losing control of data, permissions or integrations.
Jul 2026Building a SaaS · 1
08 / LET'S TALK
Working through a difficult AI or product problem?
Tell me what you are building and where it gets stuck. I’m open to senior engineering roles, product collaborations and focused consulting work.
Explore AI engineering services