“
Most frontend problems are structure problems in disguise. I settle the data model and the component contracts first, so the screens come together fast — and still make sense six months later.
01Model the data before the screenPrinciple
Driptanil Datta
Full-stack & AI Engineer
2+ years in production
2+ years in production
Based out of India
Open to remote roles and to relocating
Open to remote roles and to relocating
Currently
SDE I · Ploutos AI
SDE I · Ploutos AI
[I build enterprise data platforms in Next.js and TypeScript, and the multi-agent systems behind them. Open to full-stack, AI and ML engineering roles]
Open to rolesFull-stack, AI & ML · remote or relocation
India · ISTdriptanil.datta@gmail.com
Selected Works
An enterprise data platform, two developer tools and a shipped mobile app. Each one started with the data model, not the screen.
Datachecks Enterprise2023 — NowAI data migration platform · Ploutos AI Technologies
350+ PRs merged · 2× faster QA on the UAE Central Bank migration
Feature BookLiveDeveloper tool · Personal product
MCP server that lets Claude write docs people can keep editing
The House Hero AppShippedMobile app · Personal product
Live on Google Play and the App Store in Australia
LynxAPIOngoingDesktop tool · Personal product
Rust and Tauri desktop app · payloads never leave the machine
Experience
On one product, Datachecks, since 11.2023: first as an intern on the open-source SDK, now as SDE I leading the enterprise frontend. Waterdip Labs and Ploutos AI are the same company under its old and new name.
11.2023 — Now
Ploutos AI Technologies
Formerly Waterdip Labs · Intern → SDE I · Full-timeDatachecks, end to end: I joined as an intern on the open-source SDK and now lead the frontend of the enterprise platform built on it. The company was renamed from Waterdip Labs to Ploutos AI along the way.
→2024Vetscribe · Dice Health
Frontend Engineer · ContractAn AI medical scribe for vets: voice in, structured clinical record out.
→Capabilities
The tools behind 350+ merged pull requests at Ploutos AI and six shipped products, grouped by what they build: agents, services, web and apps.
[01]12 tools
Agentic AI
Multi-agent systems with persistent memory and retrieval, deployed on-premise when the client needs it.
- Complex agents
LangGraph
AlsoMulti-agent systems · RAG pipelines
OpenAI
Claude
Grok
Hugging Face
LangChain- LangGraph
LangSmith
Langfuse
Mem0
pgvector
PyTorch
scikit-learn
[02]17 tools
Backend
APIs and storage that hold up at 10M+ rows: schemas first, then the queries that read them.
- Small scale
Next.js- Large scale
Rust- AI services
FastAPI
AlsoREST & WebSockets · SQL · SQLite · CI/CD
Python
Node.js
Actix Web- PostgreSQL
Azure SQL
MongoDB
Redis
Snowflake
Prisma
SQLAlchemy
SeaORM
Pandas
Polars
Docker
[03]09 tools
Frontend
Enterprise web apps, built on component contracts a team can extend.
- Web apps
- Next.js
- Static sites
Astro
AlsoDesign systems
TypeScript- Next.js
React- Astro
Tailwind CSS
shadcn/ui
Zustand
Axios
React Flow
[04]09 tools
Mobile & desktop apps
Flutter apps shipped to Google Play and the App Store, and Rust and Tauri desktop apps that keep payloads on the user's machine.
- Mobile apps
Flutter- Desktop apps
Tauri
AlsoApp Store & Google Play releases · OpenAPI
- Flutter
Dart
Material
Cupertino
GetX- Dio
Firebase Messaging- Tauri
- Rust
2,063 Contributions
My GitHub activity over the 12 months to Oct 2026. 2,056 of them are in private work repositories, which GitHub counts but doesn't name.
OctNovDecJanFebMarAprMayJunJulAugSeptOctMonWedFri
Hover a day for its countLess More
285Active days, last 12 months
34Longest streak, in days
33Busiest day, 15 Jul
28Public repos since 2019
Desk Setup
Where the work happens. Late nights, a big screen and a terminal always open, with the lighting changing as the day goes. Click a photo to see it full size.
On the desk
- MacBook Pro
- An external display
- A light bar on the display
- Keyboard and mouse
- Lighting that follows the hour
NOW LEARNING
A deep understanding of Deep Learning & Neural Networks with Python
42% · 25 of 59 lectures
Next up: Overfitting and Cross-Validation.
Gradient Descent5/5
Artificial Neural Networks (ANNs)5/5
Overfitting and Cross-Validation0/4
Regularization0/4
Feed-Forward Networks (FFNs) & Milestone Projects0/4
Autoencoders0/4
5 certificates earned
Each one opens full size and links to Udemy's verification page, so anyone can check the credential.
Notes
396 published pages across 21 series. I write while I learn, from the terminal up to LLM internals. Most recent first.
How I Work
Four principles I hold myself to, and three things a team gets from them. Scroll sideways for all seven.
“
State that lives in one place, components that own one job, and side effects you can point at. Boundaries are what let a second engineer land a change without reading the whole tree.
02Make the boundaries obviousPrinciple
“
Fast local dev, honest types, and errors that say what went wrong. Cutting the loop from minutes to seconds changes what a team is willing to attempt.
03Optimise the feedback loopPrinciple
“
Shipping is the start of maintenance. I leave behind naming, docs and tests that make the next change cheap rather than clever code that makes it expensive.
04Ship it, then keep it legiblePrinciple
“
I build the boring structure well — data models, component contracts, state boundaries — so the next three features slot in instead of forcing a rewrite.
→An interface that survives the roadmapWhat you get
“
React and TypeScript on the front, FastAPI, Node and Postgres behind it, Rust when it earns its place. Fewer handoffs, fewer things lost between them.
→One engineer across the whole stackWhat you get
“
Six products in production, maintained after launch. I care about the second year of a codebase as much as the first two weeks.
→Work that ships and stays shippedWhat you get
FAQ
What teams usually ask before we start working together.
Open to full-stack, AI engineering and ML engineering roles. I'm based out of India and open to remote roles or relocating for the right team. The fastest way to reach me is driptanil.datta@gmail.com.
Most frontend problems are structure problems in disguise. I settle the data model and the component contracts first, so the screens come together fast — and still make sense six months later.
React and TypeScript on the front, FastAPI, Node and Postgres behind it, Rust when it earns its place. Fewer handoffs, fewer things lost between them.
Yes. At Ploutos AI I lead a four-person frontend team on Datachecks Enterprise. We shipped RBAC, SAML SSO, multi-workspace support, cross-database comparison, lineage and profiling.
Both. I architect multi-agent systems with persistent memory (Mem0), RAG pipelines and LangGraph orchestration, and I build the interfaces people use to run them.
Shipping is the start of maintenance. I leave behind naming, docs and tests that make the next change cheap rather than clever code that makes it expensive.