Skip to content
View gauravlochab's full-sized avatar
🎯
Focusing
🎯
Focusing
9 followers · 43 following

Achievements

Achievement: Pair Extraordinairex2Achievement: QuickdrawAchievement: Pull Sharkx2Achievement: YOLOAchievement: Arctic Code Vault Contributor

Achievements

Achievement: Pair Extraordinairex2Achievement: QuickdrawAchievement: Pull Sharkx2Achievement: YOLOAchievement: Arctic Code Vault Contributor

Block or report gauravlochab

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
gauravlochab/README.md
   

🧠 About Me

I build autonomous agent systems and production LLM pipelines — from fine-tuning language models at scale to shipping full-stack AI applications used in real businesses.
🔭 Currently building Agentic platforms with LangGraph, CrewAI & Claude Agent SDK
🤖 Specialising in Multi-agent orchestration, LLM fine-tuning, NLP at scale
🌏 Based in Delhi, India
💼 Open to Consulting, contracts, and AI collaboration
📬 Contact Via GitHub or LinkedIn

🚀 Featured Projects

🧾 Agentic Invoice Processor

Production AI agent platform for automated invoice processing and compliance workflows.

  • LangGraph 7-node state machine with human-in-the-loop interrupt gates
  • Multi-model GPT-4-turbo extraction → GPT-4o contract matching
  • NL chat — approve invoices in plain English via LangChain tool-calling
  • Email ingestion, OCR, audit trail, React dashboard

LangGraph FastAPI OpenAI React PostgreSQL

🧪 LLM Fine-Tuning Pipeline

QLoRA fine-tuning of Llama 3.1 & Qwen 2.5 for structured information extraction.

  • 99.3% JSON validity, 91.6% field accuracy on Llama 3.1 8B
  • QLoRA — 4-bit NF4, rank-8 LoRA, 0.46% trainable params (28GB → 4.5GB)
  • SPLADE sparse embeddings server for Elasticsearch-native semantic search
  • 2.7M+ documents processed end-to-end

QLoRA Llama 3.1 Qwen 2.5 vLLM SPLADE PyTorch

🎙️ AI Mock Interview System

Multi-agent technical interview platform with real-time scoring.

  • 4 specialised async agents: Orchestrator, Interviewer, Evaluator, Topic Manager
  • Weighted scoring: Technical 30% / Depth 25% / Communication 25% / Application 20%
  • Dynamic topic transitions based on live performance
  • Supports OpenAI GPT-4o & Anthropic Claude

Anthropic OpenAI Multi-Agent Python AsyncIO

⚡ Optimus — Valory

Autonomous on-chain agent built at Valory — executes DeFi strategies end-to-end without human intervention.

  • Autonomous agent — perceives chain state, reasons, and executes transactions independently
  • Built on Open Autonomy framework for decentralised agent services
  • Multi-agent coordination with consensus and fault tolerance
  • Deployed on-chain as part of the Olas ecosystem

Open Autonomy Python On-Chain Agents DeFi Olas

🛠️ Tech Stack

AI & LLM

LangChain OpenAI Anthropic HuggingFace PyTorch vLLM

Backend & Infra

Python FastAPI PostgreSQL Redis Docker Celery

Frontend

TypeScript React Next.js Tailwind

📊 2026 Contributions


Jan 1 – Apr 5, 2026
Projected ~1,970 for full year

⭐⭐⭐⭐☆

Activity Rating (4 / 5)

Top Languages

Python TypeScript JavaScript Solidity Shell

Pinned Loading

  1. agentic-invoice-processor agentic-invoice-processor Public

    Production AI agent platform: LangGraph state machine + GPT-4 for automated invoice processing, contract matching, and compliance workflows

    Python

  2. llm-finetuning-structured-extraction llm-finetuning-structured-extraction Public

    QLoRA fine-tuning pipeline: Llama 3.1 & Qwen 2.5 for structured extraction — SPLADE sparse embeddings, vLLM FP8 inference, end-to-end data pipeline

    Python

  3. optimus optimus Public

    Forked from valory-xyz/optimus

    Python

  4. autonolas-tokenomics-subgraph autonolas-tokenomics-subgraph Public

    Forked from valory-xyz/autonolas-subgraph-studio

    Subgraph for Autonolas tokenomics contracts

    TypeScript

  5. RAG-Private-GPT-Langchain RAG-Private-GPT-Langchain Public

    This is the implementation of Private-GPT in Langchain

    Python 1

  6. arya-website arya-website Public

    Personal website and technical portfolio

    TypeScript