Tejus Gupta

AI Engineer focused on agents, developer tooling, and intelligent systems

Software development is an art that requires technical depth — not a stressful job, but a craft I genuinely enjoy practicing.

Philosophy

I enjoy building things and understanding how they work at a low level. I like breaking large systems into smaller, solvable problems and making them work reliably. I especially enjoy building tools that remove friction from everyday developer workflows.

When I build, I value quality above everything — in code, in user experience, and in long-term maintainability. I care deeply about clean setups, automation, and making things feel "done", not just "working".

I'm most excited by recurring, real-world problems — the kind where writing a tool once can save hours every day.

My Work

create-agentverse-agent

A CLI tool that scaffolds production-ready uAgents projects with Agentverse-compatible defaults.

Why I built it

I was repeatedly writing the same setup code when creating new agents. This tool removes that friction and provides a clean, consistent starting point for serious agent development.

What makes it interesting

Generates production-focused scaffolds, not demo templates. Built-in async patterns, logging, health checks, and quota handling. Designed around real agent development workflows.

Tech Stack

Python · CLI tooling · Jinja templating · Async-first design · Docker & Docker Compose · Agentverse / uAgents ecosystem

Bhasha

A toy programming language interpreter using Hindi keywords.

Why I built it

To explore language design and interpreter internals while experimenting with non-English syntax and semantics.

What makes it interesting

Covers core interpreter concepts: parsing, evaluation, execution. Uses Hindi keywords for control flow and expressions. A hands-on way to understand how languages actually work.

Tech Stack

Rust

PassCrypt

A secure, CLI-based password manager.

Why I built it

I wanted a reliable, portable password manager that fits naturally into a terminal-centric workflow.

What makes it interesting

Uses strong symmetric encryption (Fernet). Supports password generation, import/export, and master password protection. Comes with a dedicated documentation site.

Tech Stack

Python · Cryptography (Fernet) · CLI tooling · Sphinx documentation

bozo

Production-grade structured logging for Python with an optional real-time web viewer.

Why I built it

Every Python service ends up rebuilding the same logging stack from scratch. bozo is the one-call setup that gets console formatting, file rotation, JSON output, and context propagation right by default — so projects ship with proper observability from day one.

What makes it interesting

One bozo.setup() wires up colored console, rotating text files, structured JSON, HTTP shipping, and an optional real-time web dashboard simultaneously. bozo.redact() handles secrets, bozo.log_context() survives async hops, and old runs are pruned automatically. Fully type-safe on Python 3.13+.

Tech Stack

Python 3.13+ · Structured logging · Async / threaded context · File rotation · Optional FastAPI-based web viewer · MIT

Vox

Connect a PostgreSQL database, ask questions in plain English, watch SQL and results stream back live.

Why I built it

Most "ask your data" tools either dump answers in a black box or hand you a half-broken query. I wanted the SQL itself to be the artifact — visible, editable, and guaranteed read-only — so it works as both an answer and a learning surface.

What makes it interesting

SQL streams token-by-token over Server-Sent Events, then result rows stream in as the query runs — the table starts filling before the LLM has finished thinking. Schema is introspected, the LLM provider is configurable, and auth is handled by Supabase (GitHub OAuth).

Tech Stack

TypeScript · Next.js 16 · Supabase · LangChain · LlamaIndex · SSE streaming · Tailwind · Framer Motion

FlowNet

A decentralized delivery network powered by autonomous AI agents.

Why I built it

To experiment with multi-agent systems, negotiation workflows, and real-world automation problems like logistics and coordination.

What makes it interesting

Combines AI assistants with autonomous agents. Designed for decentralized, multi-agent interaction. Bridges AI reasoning with marketplace-style workflows.

Tech Stack

Python · AI / LLMs · Async networking · API servers · Databases and queues

uagents-composio-adapter

A production-ready adapter connecting uAgents with Composio and LangChain ecosystems.

Why I built it

To enable full async integration between uAgents, Composio tools, and LangChain-based agents — including authentication and orchestration.

What makes it interesting

Brings multiple agent ecosystems together. Async-first, orchestration-focused design. Actively versioned and production-oriented.

Tech Stack

Python (async) · uAgents · LangChain · Composio APIs

How I Build

For any new project, I start with a solid foundation: CI/CD, testing, linting, formatting, and pre-commit hooks — before writing application logic. Once the foundation is ready, I dive deep into research, explore similar solutions, choose the right tools, and define a clear roadmap.

My priorities are simple:

  1. It must work correctly
  2. It should be performant
  3. It should be readable
  4. It must be well-documented

A non-negotiable for me is user experience — every tool I build aims to feel easy and natural to use.

Now

Currently learning

Building a peer-to-peer relay layer for CLI tools in Rust using custom protocols.

Currently experimenting with

Building CLI tools to automate daily workflows and fine-tuning my Linux environment with custom aliases, scripts, and shortcuts.

Work

Currently working as an AI Engineer at Fetch.ai, building AI agents using frameworks like Google ADK, OpenAI Agents SDK, LangChain, and CrewAI.