Biography & Story

About Jeet

Agentic Engineer & data enthusiast based in Bangladesh. Bridging the gap between frontier AI reasoning and scalable, real-world data pipelines.

The Mission

I specialize in designing and implementing Agentic AI architectures — software where models do not just chat, but autonomously plan, orchestrate tasks, execute code, and query heterogeneous data stores.

My journey began in deep data pipelines, wrangling high-throughput streams and optimizing analytical databases. When large language models unlocked reasoning capabilities, combining robust data infrastructure with autonomous agent systems became my core obsession.

When I'm not architecting multi-agent loops or tuning vector stores, you can find me experimenting with modern frontend interfaces, digging into open-source repositories, or writing about AI engineering.

2+
Years Experience
Engineering AI agents & data pipelines
10+
Projects Shipped
From multi-agent swarms to stream engines
50K/s
Peak Event Throughput
Real-time Kafka & streaming architectures
15+
Core Technologies
Python, Next.js, LangChain, Kafka, Docker

Skills & Tech Stack

Frontend
ReactNext.jsTypeScript
Tailwind CSSHTML/CSS
Backend & Data
PythonNode.jsPostgreSQL
MongoDBFastAPI
AI & Tools
LangChainOpenAIDocker
GitLinux

Experience Journey

2024 — Present

Agentic Systems & AI Tool Execution

Designing autonomous multi-agent swarms with tool use, persistent memories, and deterministic fallback loops. Focusing on enterprise RAG and production LLM orchestration.

2023 — 2024

Data Streaming & Event Architectures

Built distributed event streaming pipelines processing telemetry data using Apache Kafka, PostgreSQL, ClickHouse, and Python microservices.

2022 — 2023

Full-Stack Software Engineering

Developed web applications, modern RESTful and WebSocket APIs, and interactive frontend interfaces using React, Next.js, TypeScript, and Node.js.

Engineering Philosophy

Deterministic Tool Execution
LLMs provide reasoning; our deterministic tools provide reliable actions. Reliable agents require strict schema validation and predictable fail-safes.
Data Freshness & Integrity
An agent is only as good as the context it retrieves. Fast streaming ingestion and hybrid dense-sparse vector indexing prevent outdated context.
Simplicity in Architecture
Composable, modular services beat bloated monolithic frameworks every time. Build primitives that do one job exceptionally well.

Interested in collaborating?

Check out my projects or get in touch for new opportunities and consulting.