PLAYER 1 READY

SOURAV
ROY

data-ai architect ◆ shipping pipelines

Qlik Certified Solution Architect building the full data-AI stack — enterprise ETL/ELT on Talend, Python, Snowflake, BigQuery, AWS & GCP, and LLM-powered pipelines.

see the work→ let's connect↗

CHARACTER FILE

TAP ME

ETL/ELT

MAX

SNOWFLAKE

PYTHON

AWS + GCP

LLMs

LEVEL 01 · IDENTITY

who / what

sourav@data-ai ~ /about

sourav@data-ai $ whoami

Data engineer building the pipes that fintech quietly runs on. By day, I ship production ETL/ELT systems where dependability matters most - pipelines you notice only when they break, and mine rarely do. By night, I fall down Python rabbit holes and wonder what LLMs make possible next.

sourav@data-ai $ cat philosophy.md

I care about systems that just work. Warehouses that answer questions instead of raising them. Audit trails that hold up under scrutiny. Code that saves someone's afternoon.

sourav@data-ai $ ls -la certifications/

# Qlik Certified Solution Architect

sourav@data-ai $

LEVEL 02 · INVENTORY

the stack

◈ CORE

Talend Snowflake Python SQL PL/SQL

☁ CLOUD

AWS GCP Azure

⊕ DATABASES

PostgreSQL Cassandra MongoDB Oracle DuckDB

✦ AI / LLM

LLM Pipelines Generative AI Vector DBs

⚙ TOOLING

Java Maven Docker Git Linux Shell

◎ EXPLORING

Video automation Blender scripting DuckDB internals

LEVEL 03 · CURRENT QUEST

what I'm building

slothDB

Fast & STEADY · C++ & Vectorization

SlothDB

C++ & Vectorization, thinking out loud

A data project I'm building and exploring what lightweight, embedded analytics can do when you give it room to breathe. Equal parts experiment and love letter to DuckDB's design.

C++ Vectorization Physical Hashing CSV, Arrow, Parquet, etc

view on github →

EXPLORING

LLM-Powered Pipelines

Injecting language models into data pipelines — classification, extraction, enrichment. Early-stage experimentation on where LLMs genuinely earn their compute in ETL.

PRODUCTION

Enterprise ETL Systems

Talend-based ingestion and transformation pipelines moving transaction data at fintech scale — with audit trails, rollback paths, and reconciliation baked in.

TINKERING