Open to work · Mahesana, India

Gaurav Sharma AI Engineer · Generative & Agentic AI

I design and ship AI systems that move beyond prototypes — generative assistants, autonomous agents and computer vision pipelines built for production.

View projects Download Resume

01 — About

I build AI systems that solve real-world problems.

My primary interests are Generative AI, Agentic AI, Computer Vision, Machine Learning and Linux systems. I enjoy designing AI applications that move beyond prototypes into production-ready software.

My experience includes developing intelligent assistants, computer vision applications, workflow automation and end-to-end AI solutions using Python and modern AI frameworks. I continuously learn by building — and I'm always interested in collaborating on impactful AI products and innovative startups.

3+Years building
6+AI projects shipped
RAGRetrieval systems
CVVision pipelines

02 — Skills

What I work with

Generative AI

LLM applications, RAG pipelines, prompt engineering, embeddings and semantic search.

  • LangChain
  • Llama
  • Hugging Face
  • ChromaDB

Agentic AI

Tool-calling agents, planning loops and autonomous workflows with human approval gates.

  • AI Agents
  • Tool calling
  • Orchestration

Computer Vision

Detection, OCR and end-to-end recognition pipelines optimised for real conditions.

  • OpenCV
  • YOLOv8
  • PyTorch

Machine Learning

Model training, evaluation and deployment with reproducible experiment workflows.

  • PyTorch
  • scikit-learn
  • MLOps basics

Engineering

Python services, data pipelines and APIs designed for reliability and scale.

  • Python
  • MySQL
  • Vector DBs

Systems & Tooling

Linux-first development, containerisation and clean version-controlled delivery.

  • Linux
  • Docker
  • Git

03 — Experience

A timeline of building

  1. Jan 2025 — Mar 2026 · Remote

    Founder — Atlas Intelligence AI

    Founded and managed an AI consulting company delivering custom AI and automation solutions: client acquisition, project scoping and technical execution. Designed LLM and ML applications that automate real business workflows, and led end-to-end operations from proposal to delivery.

  2. Jan 2026 — Feb 2026 · Remote

    Generative AI Intern — Carrier Complier

    Developed a Retrieval-Augmented Generation chatbot with LangChain and Llama over indexed educational documents. Built the full retrieval pipeline — ingestion, chunking, embeddings, vector search — improving answer accuracy by grounding outputs in source content.

  3. Apr 2025 — Sep 2025 · Remote

    Data Science Intern — The Entrepreneurship Network

    Designed and developed NexusClaw, a full-stack AI agent platform integrating LLMs, RAG and tool-calling workflows behind an interactive web interface. Built RAG-based FAQ chatbots for educational content.

  4. Oct 2023 — Sep 2025 · Hybrid

    Mathematics & Computer Science Educator — Self Employed

    Taught mathematics (Classes 6–10) and computer science (Classes 11–12), covering programming fundamentals, algorithms, data structures and databases with personalised lesson plans and assessments.

Education

Dr. A.P.J. Abdul Kalam Technical University

B.Tech, Computer Science & Engineering · Sep 2022 — May 2026

  • Grade 6.24

Central Board of Secondary Education

10 + 2 · Apr 2015 — Jul 2022

  • 80% & 75%

04 — Featured projects

Selected work

AI Multi-Research Agent

A four-stage multi-agent research pipeline (Search, Reader, Writer, Critic) using LangChain and Gemini 2.5 Flash, automating end-to-end report generation from a single topic query, with a Streamlit interface and a standalone CLI runner.

  • LangChain
  • Gemini
  • Streamlit
  • Python
GitHub Live demo

AI Legal Document Analyzer

A fully local, privacy-preserving RAG system (LangChain, Ollama, ChromaDB) for legal summarisation, semantic search and grounded Q&A with zero third-party API calls. Similarity-threshold gating prevents hallucinated answers.

  • Ollama
  • ChromaDB
  • RAG
  • Python
GitHub

NexusClaw

An agentic CLI tool (TypeScript, Bun, Vercel AI SDK) with an in-memory staging layer that queues file mutations for review — model-agnostic routing via OpenRouter, plus Telegram-based approval of agent-generated diffs.

  • TypeScript
  • Bun
  • AI SDK
  • OpenRouter
GitHub

Knight-Sight (ANPR Pipeline)

End-to-end automatic number plate recognition — vehicle detection, plate localisation and OCR — using a custom-trained YOLOv8 model. CLAHE-based low-light enhancement and adaptive-threshold glare mitigation boost recognition reliability in real conditions.

  • YOLOv8
  • OpenCV
  • OCR
  • Streamlit
GitHub Live demo

05 — Certifications

Licenses & certifications

AI Fluency for Small Businesses

Anthropic · Issued Aug 2026

  • Credential j3p8dex75k9k

Certificate of Participation

Bennett University · Issued Apr 2026

  • Hackathon

Prompt Engineering & AI Agents with ChatGPT

LinkedIn · Issued Jul 2026

  • Agents

06 — Tech stack

Daily tools

Python
PyTorch
LangChain
Hugging Face
OpenCV
Vector DBs
ChromaDB
Ollama
MySQL
Docker
Linux
Git
Streamlit
TypeScript

07 — Contact

Let's build something intelligent.

Open to AI engineering roles and collaborations on impactful AI products.

Get in touch

The fastest way to reach me is email — I usually reply within a day.

chmodgaurav@gmail.com Connect on LinkedIn