Architecture & Scoping Sprint
From $1,500- Requirements and user flows
- System design and stack choice
- Build plan with cost estimate
Good for: founders before a build, teams before a rewrite
Taking remote projects · IST, overlaps EU & US mornings
Full-stack products with AI agents built in, shipped to production.
I help SaaS and dev-tools startups ship web apps, APIs and AI agents that hold up in production. Principal Engineer and AI Architect, 20 years of building platforms, and I write the code myself.
Fastest path: one call, then a written scope and fixed quote within 48 hours.
POST /api/v1/assistant/asktrace 7f3a·c21e
webReact · Vite842ms
api gatewayNestJS · auth791ms
policyOPA · Rego31ms
retrieverpgvector · rerank142ms
agentADK · planner470ms
tool callMCP · Jira186ms
cacheRedis4ms
200 OK · eval passk8s · OpenTelemetry
Full-stack web & mobile
Product UIs, admin consoles and APIs with auth, payments and roles done properly.
Agentic AI & RAG
Multi-agent workflows, tool use over MCP, retrieval, guardrails and evaluation harnesses.
Cloud & platform
Containerised deployments, CI/CD, observability and cost control on any of the big three clouds.
Ways to work together
Freelance engagements
Fixed-scope packages for clear problems, or a monthly retainer when you need a senior engineer on call. All remote, with written updates you can forward to stakeholders. Prices are starting points in USD; you get a fixed quote after the scoping call.
Free · 30 minutes
Book the free review
AI readiness review
Bring your product and one workflow. You leave with two or three places AI would genuinely help, a rough build cost, and the risks to watch. No pitch deck, no obligation.
MVP Build
From $8,000- Web or mobile app with backend API
- Auth, roles, admin console
- CI/CD and cloud deployment
Good for: a working product in users' hands
AI Agent & RAG Integration
From $5,000- Agents wired to your tools via MCP
- Retrieval over your documents and data
- Evaluation set and regression tests
Good for: adding AI to an existing product
Fractional Tech Lead
From $2,500 / month- Weekly architecture and code review
- Vendor and model selection
- Hands-on fixes when they matter
Good for: small teams without a senior architect
Technical guidance for startups
A senior technical voice before you can hire one
Early technical choices are cheap to make and expensive to undo. I help non-technical founders and small teams make them with a clear head, through short advisory calls or a standing monthly slot.
Idea
Validate cheaply
Clickable prototype or no-code test before writing real code. Decide what must be custom and what can be bought.
MVP
Build the smallest real thing
One stack, one cloud, managed services, auth and payments from proven providers. Ship in weeks, not quarters.
Pre-seed / Seed
Make it dependable
CI/CD, backups, monitoring and basic security so demos don't break in front of investors or first customers.
Series A
Scale the team
Architecture that new engineers can learn, cost controls, compliance groundwork and a hiring plan for the first tech lead.
Stack & architecture choice
A recommendation based on your product, budget and the developers you can realistically hire in your market.
You get: a one-page architecture decision with trade-offs
AI strategy & build vs. buy
Where an LLM or agent actually helps your users, which model and vendor to start with, and how to keep token costs predictable.
You get: ranked AI use cases with cost estimates
Vendor & quote review
An independent read of proposals, estimates and contracts from agencies or freelancers before you sign.
You get: red flags, missing items and questions to ask
Cloud cost & scalability
Review of your Azure, GCP or AWS setup for waste, right-sizing and the next bottleneck you'll hit as users grow.
You get: a prioritised savings and scaling list
Code & tech due diligence
An honest health check of a codebase for founders taking over a product, or investors evaluating one.
You get: a risk report with fix-now and fix-later items
Hiring & team setup
Role definitions, interview tasks and help assessing candidates, plus the engineering practices to set up before they join.
You get: job specs and a technical interview kit
Use cases
Typical problems I solve
Example scenarios showing how a request turns into a build. Each one draws on systems I've already shipped.
B2B SaaS
Support copilot over your docs
The problem
Support tickets keep asking questions already answered in docs, past tickets and release notes.
What I build
A RAG assistant inside your app and helpdesk, with source citations, hand-off to a human, and an eval set that catches wrong answers before release.
Operations teams
Agent that runs routine workflows
The problem
Staff spend hours each week pulling status from Jira, CRM or spreadsheets and writing the same reports.
What I build
An agent connected to your tools over MCP that answers questions, drafts reports and runs approved multi-step actions, with policy checks on every tool call.
Health & consumer apps
Mobile app with web admin
The problem
A founder needs iOS and Android apps, a customer web app and an admin console, without three separate teams.
What I build
One Flutter codebase for mobile, a React web and admin app, and a single API with subscriptions, notifications and device integrations.
Low-code teams
Power Apps that need a real backend
The problem
A PowerApps solution has outgrown its data limits, gets slow, and can't do search or AI.
What I build
A Python or Node API on Azure behind a custom connector, so the familiar front end stays while search, AI and scale move to the backend.
Media & content
Content platform with roles
The problem
Editors, partners and admins share one messy back office, and every new client wants their own branding.
What I build
A multi-tenant CMS with role-based access, white-label theming, audit logs and an API that feeds web, mobile and TV clients.
Selected work
Things I've designed and built
A mix of enterprise work and products I've built as a founder. Client names are left out where the work is under NDA.
Agentic AI · Enterprise2024 → now
Release intelligence copilot
A multi-agent assistant for release managers at a global semiconductor software company. It answers questions about releases and runs multi-step workflows across Jira and internal systems.
Built withGoogle ADK sequential, parallel and loop agents · multiple MCP servers · OPA/Rego policy checks · Redis-backed memory · SSE streaming API · golden-dataset eval harness
Platform · Enterprise2024 → now
Unified API & MCP hub
One API and MCP layer over a set of SDLC and DevOps tools, so internal apps and AI agents use a single, governed interface instead of a dozen integrations.
Built withAPI gateway patterns · MCP · data normalisation pipelines · cloud-native deployment
Health-tech · Founder buildProduct
DropSugar
A metabolic health platform: glucose tracking (manual and CGM), nutrition, activity, medications, AI coaching, community challenges, telehealth booking and subscriptions.
Built withNestJS · Prisma · React + Vite web and admin apps · Flutter iOS/Android · Docker · GitHub Actions · Playwright end-to-end tests
HR-tech · Full stackProduct
TalentSpotter
A recruitment platform with resume intelligence and a RAG chat for recruiters, plus a low-code front end so business teams can run searches themselves.
Built withPython Flask · Azure Functions · Cosmos DB · PowerApps custom connectors · OpenAPI · Power Automate
Media · ArchitectureYashodhanTek
Streaming CMS platform
A control-plane CMS for OTT businesses, with an operator portal and clients for mobile, TV and the web. Covers DRM, entitlements and white-label branding.
Built withNode.js / TypeScript services · OIDC · RBAC · event-driven outbox · Tailwind operator portal · Kubernetes autoscaling
AI platform · DesignReusable
Multi-tenant RAG platform
A reusable, Kubernetes-hosted retrieval application that teams can deploy for their own document sets, with tenant isolation from ingestion to chat.
Built withIngestion pipelines · vector search · caching · monitoring · chat UI · per-tenant config
About
Senior engineer, not an agency
Srinivasa Rao SHyderabad, India · IST (UTC+5:30)
20
years building software
3
clouds: Azure, GCP, AWS
PoC→GA
AI agents taken to production
MBA
Data Science, 2024–26
I'm Srinivasa Rao, a Principal Engineer and AI Architect based in Hyderabad. My day job is building agentic AI and platform tooling at a global semiconductor software company. Before that I spent years in platform engineering, DevOps and enterprise architecture, including disaster-recovery programmes for large US companies.
That mix is the point. I can build the React screen, the API behind it and the agent behind that, then deploy it with monitoring, policies and cost controls in place. You work with me directly from first call to handover.
- 2024 → nowPrincipal Engineer · AI ArchitectAgentic AI and API platforms, global semiconductor software company
- 2023 → 2024Contract engineer, same companyDevOps tooling and early agent prototypes
- OngoingFounder engineer, YashodhanTekProduct builds in streaming, health-tech and HR-tech
- 2006 → 2023Engineer to architectEnterprise platforms, cloud migration and DevOps
Toolbox
What I work with
I pick tools for your team's ability to maintain them after I leave, not for novelty.
- Frontend
- React, TypeScript, Vite, Tailwind, Next.js
- Backend
- Node.js, NestJS, Python, Flask, FastAPI, REST, SSE
- Mobile
- Flutter for iOS and Android, PowerApps
- Data
- PostgreSQL, Prisma, Cosmos DB, Redis, vector stores
- AI
- Google ADK, MCP, LiteLLM, Azure OpenAI, RAG, evaluation
- Cloud
- Azure, GCP / Vertex AI, AWS, serverless functions
- Platform
- Docker, Kubernetes, Terraform, GitHub Actions, Argo CD
- Quality
- Playwright, OpenTelemetry, OPA policies, load tests
How a project runs
Three stages, in this order
Week 0
Scope
A call to understand goals, users and constraints. You get a written scope, architecture sketch, timeline and fixed quote.
Weeks 1–N
Build
Weekly demos on a staging link. Code lives in your repository from day one, with tests and CI running.
Final week
Harden & hand over
Monitoring, load checks and security review, then docs and a walkthrough so your team can run it.
Writing & open source
Notes from production
Short write-ups on what actually breaks when AI agents meet real company data, plus code you can run.
Article · Evaluation
Why your agent eval passes on Monday and fails on Tuesday
Golden-dataset drift, non-deterministic tool selection, and how to make agent tests trustworthy.
Read on LinkedIn → Article · CostThe hidden token bill of exposing agents through IDEs
How silent retries multiply cost per prompt, and the guardrails to put in place before rollout.
Read on LinkedIn → Open source · MCPStarter MCP server for Jira-grounded agents
A small, documented MCP server with query validation and tests, ready to fork for your own tools.
View on GitHub →Before you ask
Common questions
How do time zones work?
I'm on IST (UTC+5:30). That gives a full working-day overlap with Europe and the Middle East, and early-morning or evening overlap with US East Coast teams. Most communication is async, with one or two scheduled calls a week.
How do you charge and invoice?
Fixed price for defined projects, paid in milestones (typically 30% to start, then on delivery of each stage). Retainers are billed monthly in advance. I invoice in USD, EUR or INR by bank transfer, Wise or Payoneer.
Who owns the code?
You do. Work happens in your repository from day one, and full IP transfers to you on payment. I'm happy to sign your NDA before we discuss details.
Can you work with our existing team?
Yes. I often join as a senior engineer alongside your developers, review their pull requests, and leave the system documented so they can run it after I step back.
What if we only need advice, not a build?
Book a single advisory session or the free AI readiness review. Plenty of founders only need a second opinion on a stack choice or an agency quote.
How soon can you start?
Usually within one to two weeks of a signed scope. Small advisory sessions can often happen the same week.
Have a project to scope?
Pick a 30-minute slot that suits you, or send a few lines about what you're building, your timeline and budget range. I reply to email within one working day.