Vaibhav Barman

AI Engineer | LLM Applications, RAG & AI Agents

I build AI agents that check their own work: LLM apps, RAG pipelines and multi-agent systems, often on local models.

ProjectsResumeGitHub LinkedIn

based in Jaipur, Rajasthan, India

vaibhav@portfolio:~ `
vaibhav@portfolio:~$ whoami
Vaibhav Barman · AI Engineer
I build AI agents that check their own work: LLM apps, RAG pipelines and multi-agent systems, often on local models.
status: open to remote roles
next: projects · skills · neofetch · help

about.md

Building AI that ships.

Portrait of Vaibhav Barman

AI engineer building LLM applications, RAG pipelines and multi-agent systems in Python, with a focus on local models, structured outputs and evaluation. I study Computer Science at BITS Pilani (B.Sc., expected 2027) and learn by shipping: ten open-source projects since August 2026.

location
Jaipur, Rajasthan, India
studying
B.Sc. in Computer Science · BITS Pilani
speaks
English, Hindi
status
open to remote roles

Focus areas

  • LLM Engineering and AI AgentsBuilding LLM-powered applications, tool-using agents, structured prompting workflows, and systems that evaluate or improve their own outputs.
  • RAG and AI SystemsExploring document ingestion, embeddings, semantic retrieval, grounded generation, and practical ways to evaluate retrieval quality.
  • Local-First AIExperimenting with locally hosted models and privacy-focused AI workflows that reduce reliance on cloud inference.

Off-screen

  • Gym and FitnessI enjoy strength training, staying consistent with fitness, and improving my performance over time.
  • Marathon RunningI enjoy challenging myself through running and endurance events. Race medals: 🥇 10 km : Run for Humanity; 🥈 5 km : JECRC's Martyr's Run.
  • Books and ReadingI enjoy reading books to explore new ideas, learn continuously, and broaden my perspective.
  • PianoI enjoy piano and the process of learning music through regular practice.
  • 10 km · Run for Humanity
  • 5 km · JECRC's Martyr's Run

~/projects · 10 entries

Things I've built.

Agentic and AI-first work first. Hover a featured project to see how it works.

  1. 01Agentic AIfeaturedshipped Oct 2026

    GitCommit AI

    An agentic, repository-aware Git commit assistant (git ai-commit) that learns how a repository writes commits and drafts messages in that style, checked by a deterministic critic before you see them.

    • 6LLM providers behind one protocol
    • 2-stagecritic: rules, then faithfulness
    • offlineeval mode, no API key
    • Detects the repository's commit convention and style profile from recent history, with a confidence score.
    • A critic loop checks each draft against deterministic rules plus an LLM faithfulness check, and regenerates until it passes.
    2 more
    • Never stages files and refuses to commit if the staging area or HEAD changed while it was working.
    • Supports Anthropic, Gemini, Hugging Face, OpenAI, OpenRouter and Ollama (local or remote), with an evaluation harness that replays your own history.
    • LangGraph
    • Python
    • Multi-agent
    • OpenAI
    • Anthropic
    • Gemini
    • Ollama

    How it works : GitCommit AIRepository for GitCommit AI (opens in a new tab)

  2. 02Agentic AIfeaturedshipped Oct 2026

    LocalForge-AI

    An autonomous, local-first multi-agent software development and QA engine built with LangGraph and Ollama. It coordinates task specification, code generation, execution, and repair workflows on a local machine.

    • 3role-specialised agents
    • 0network calls at runtime
    • pytestdeterministic harness
    • Uses specialized agents for specification, implementation, and verification stages.
    • Feeds concrete execution, linting, and test failures back into a code-repair loop.
    1 more
    • Designed for offline runtime with locally hosted open-weight models through Ollama.
    • LangGraph
    • Ollama
    • Python
    • Multi-agent
    • Local LLMs

    How it works : LocalForge-AIRepository for LocalForge-AI (opens in a new tab)

  3. 03Agentic AIfeaturedshipped Aug 2026

    AI Payment Recovery Agent

    An AI agent prototype that classifies failed payments, selects a recovery strategy, simulates outcomes, and adapts later decisions using historical performance.

    • 6payment-failure types modelled
    • ≤3attempts per transaction
    • JSONdiagnosis, strategy, confidence
    • Uses LLM reasoning to classify payment-failure context and choose a recovery action.
    • Records simulated outcomes and feeds strategy-performance history into subsequent decisions.
    1 more
    • Combines a web dashboard with a FastAPI backend, SQLite, and a Groq-hosted LLM.
    • AI Agents
    • FastAPI
    • SQLite
    • Groq

    How it works : AI Payment Recovery AgentRepository for AI Payment Recovery Agent (opens in a new tab)

  4. 04RAGfeaturedshipped Sep 2026

    Odyssey RAG

    A retrieval-augmented question-answering system that answers questions about Homer's The Odyssey using relevant passages retrieved from the source text.

    • 850chunks (1,000 chars, 200 overlap)
    • 768-dGemini embeddings
    • top-5cosine retrieval, Recall@K + MRR
    • Processes a source document into overlapping chunks and retrieves relevant passages using embedding similarity.
    • Uses Gemini embeddings and Gemini for grounded answer generation.
    1 more
    • Includes retrieval inspection and evaluation with metrics such as Recall@K and Mean Reciprocal Rank.
    • RAG
    • Gemini
    • Embeddings
    • Evaluation
    • Python

    How it works : Odyssey RAGRepository for Odyssey RAG (opens in a new tab)

  5. 05LLM appshipped Oct 2026

    PromptRoute

    An LLM-native travel itinerary planner that generates structured, time-slotted travel plans using a local Ollama model and an interactive web interface.

    • Explores a Backend-as-a-Prompt architecture that returns structured JSON without a conventional application server or database.
    • Connects local model output to an interactive web frontend.
    1 more
    • Designed for local inference without cloud API keys or per-token inference costs.
    • Ollama
    • Local LLMs
    • Structured output

    Repository for PromptRoute (opens in a new tab)

  6. 06LLM appshipped Oct 2026

    Local YouTube Video Summarizer

    A privacy-focused tool that transcribes YouTube video audio and creates concise summaries using Whisper and a locally hosted language model through Ollama.

    • Uses Whisper for audio transcription and Ollama for local LLM summarization.
    • Provides command-line and Gradio web-interface workflows.
    1 more
    • Keeps transcription and summarization on the user's own machine.
    • Whisper
    • Ollama
    • Gradio
    • Python

    Repository for Local YouTube Video Summarizer (opens in a new tab)

  7. 07LLM appshipped Oct 2026

    ApplyIQ

    An AI-powered job application intelligence platform that compares a candidate's resume with a job description to surface ATS compatibility, missing skills, and potential recruiter concerns.

    • Integrates Gemini-based resume and job-description analysis into a full-stack application.
    • Pairs a web frontend with a Python FastAPI backend.
    1 more
    • Includes PostgreSQL persistence with SQLAlchemy and Alembic, input validation, prompt-injection defenses, and automated tests.
    • Gemini
    • FastAPI
    • PostgreSQL
    • SQLAlchemy
    • Alembic

    Repository for ApplyIQ (opens in a new tab)

  8. 08LLM appshipped Sep 2026

    Gemini Website Summarizer

    A Python application that extracts readable content from a webpage and uses Google's Gemini API to generate a concise Markdown summary.

    • Fetches webpages with Requests and parses readable content with BeautifulSoup.
    • Uses the Google Gen AI SDK to generate summaries.
    1 more
    • Loads API credentials from local environment configuration.
    • Gemini
    • BeautifulSoup
    • Python

    Repository for Gemini Website Summarizer (opens in a new tab)

  9. 09AI toolingshipped Oct 2026

    GNews MCP Server

    A Model Context Protocol server that exposes news search and top-headline retrieval tools for AI assistants.

    • Implements MCP tools for searching news articles and retrieving top headlines.
    • Supports configurable language, country, category, date range, sorting, and result limits.
    1 more
    • Built in Python with FastMCP and managed with uv.
    • MCP
    • FastMCP
    • Python
    • uv

    Repository for GNews MCP Server (opens in a new tab)

  10. 10Web appshipped Nov 2026

    Spendly

    A lightweight personal expense tracker built with Flask and SQLite, developed end to end with Claude Code as an agentic coding workflow.

    • Server-rendered Flask app with Jinja2 templates and a shared base layout.
    • SQLite persistence with parameterized queries kept in a single database module.
    2 more
    • Built with Claude Code using a project CLAUDE.md, custom skills and subagents for exploration and test verification.
    • Tested with pytest and pytest-flask.
    • Flask
    • SQLite
    • Python
    • Claude Code
    • pytest

    Repository for Spendly (opens in a new tab)

skills.txt

The stack.

Grouped by category. Hover a skill to see which projects use it.

Languages & Databases

  • Python 7
  • SQL
  • MySQL
  • PostgreSQL 1
  • SQLite 2

LLM & Agent Frameworks

  • LangChain
  • LangGraph 2
  • MCP 1
  • FastMCP 1
  • RAG 1
  • Local LLMs 2
  • Ollama 4

LLM Providers

  • Anthropic 1
  • OpenAI 1
  • Gemini 4
  • Groq 1
  • NVIDIA

Vector DB & Cache

  • ChromaDB
  • Redis

Backend

async APIs with auth, pagination and rate limiting

  • FastAPI 2
  • Pydantic
  • SQLAlchemy 1
  • Alembic migrations 1
  • SlowAPI rate limiting

Cloud & DevOps

  • AWS S3
  • EC2
  • Lambda
  • Boto3 Amazon's Python SDK
  • Docker
  • Nginx
  • Ubuntu Linux

Tools

  • Git
  • GitHub
  • Claude Code agentic coding 1

= number of my projects that use it

Also in my projects

  • Multi-agent
  • AI Agents
  • Embeddings
  • Evaluation
  • Whisper
  • Gradio
  • uv
  • Flask
  • pytest

experience.md · education.md

The path so far.

  1. Expected 2027

    B.Sc. in Computer Science

    BITS Pilani · India

    • CGPA 7.99/10

  2. Aug 2026

    Shipped 1 project

  3. Sep 2026

    Shipped 2 projects

  4. Oct 2026

    Shipped 6 projects

  5. Nov 2026

    Shipped 1 project

  6. next

    Your team?

    open to remote roles

contact.txt

Let's build something.

Currently open to remote roles. The fastest way to reach me:

vaibhavbarman2609@gmail.com