// full_stack_+_ai_engineer
SHANTANUGUPTA, full-stack and AI engineer
One engineer for your whole product.
I design, build and ship every layer myself, from the screen your users touch to the AI behind it, and lead the team that runs it. Senior Manager, Tech @ Eureka Forbes.
Frontend
Flutter · Next.js · Kotlin · Tauri
Backend
Node.js · TypeScript · Django · gRPC · MCP
Database
PostgreSQL · MongoDB · MySQL · SQLite · Firebase
AI
LLMs · RAG · Agents · Evals
Previously Lead App Engineer @ Eureka Forbes · Project Lead @ Designs In Change
// things_i_work_with
Technologies by layer: frontend, backend, database and AI
// one_stop_engineering
Every layer of your product. One owner.
Most products stall at the hand-offs: the app team waits on the API, the API waits on the data, and AI is bolted on last. I build all four together, so the pieces fit from day one and one person answers for the whole thing.
Frontend
Apps people use every day
Flutter apps for Android, iOS and the web, from consumer apps to field tools for technicians. Micro-frontends when one product grows into several.
- Design
- Flutter / Next.js
- State (Bloc)
- Android · iOS · Web
Done at: Aquaguard customer, technician and IoT apps
Backend
Services that hold under load
Node.js and TypeScript services, gRPC between them, and caching where the traffic peaks. Benchmarked for a million users.
- API design
- Node.js + TS
- gRPC / REST
- Cache
- Cloud
Done at: Yellowbacks, Qvenue
Database
Data in the right shape
PostgreSQL for money and location, MongoDB and Firestore for live data, and vectors when the data has to answer questions.
- Data model
- PostgreSQL
- MongoDB (live)
- Vectors
Done at: Glyde Rides, Siply Chits
AI
Models that ship, not demos
LLMs with retrieval, tools and agents, inside the evals, guardrails and cost limits that production needs.
- Your data
- RAG
- LLM + tools
- Evals
- Guardrails
Done at: AI architecture at Eureka Forbes
// things_i_solve
Bring the problem, not the spec
“I have an idea and nothing built”
FrontendBackendDatabaseAII take it from the first screen to production: the app, the API, the database, and AI where it earns its place.
Done before: Mr. Klean: user app, admin portal and delivery app, designed and built by me alone.
“My app feels slow and dated”
FrontendBackendI rebuild the screens people touch most and fix what slows them down, on the device and behind it.
Done before: The Aquaguard customer, technician and IoT apps, which I lead at Eureka Forbes.
“My backend can't keep up”
BackendDatabaseI find the hot paths, cache them, split what has to scale on its own, and load-test before your users do.
Done before: Yellowbacks: a Node.js backend I scaled and benchmarked for a million users.
“My data is scattered”
DatabaseBackendI give each kind of data the store it needs and one API in front of all of them.
Done before: Glyde Rides: MongoDB for live data, PostgreSQL for locations and transactions.
“I want AI in my product”
AIBackendDatabaseI pick the use case worth automating, ground the model in your own data, and measure it before it reaches users.
Done before: AI solution architect at Eureka Forbes, and the assistant on this site, which answers from my profile.
Something else?
Describe it to my AI assistant and see how I’d approach it, or write to me directly.
// things_i_build
Featured projects
// by_the_numbers
- years shipping software
7+
- users served by products I built
10M+
- products taken to production
10+
- enterprise brands served
6+
// previously_at
Eureka Forbes
Senior Manager, Tech2023 - nowDesigns In Change
Senior Software Engineer2021 - 23Qart Fashion
Software Developer2019 - 21Glyde Rides
Software Developer2018 - 19UrbanRider
Software Developer Intern2018 - 18
// things_i_experiment_with
AI Lab & open source
$ gh repo list --topic ai
Experiments with agents, RAG and MCP are landing here soon.
// available_for
Your whole product, end to end
An idea that needs building from zero, or one layer that’s holding the rest back: frontend, backend, database or AI. I also take on solution architecture and engineering leadership.