Skip to content
View pras-ops's full-sized avatar
3 followers · 1 following

Achievements

Achievement: YOLOAchievement: StarstruckAchievement: Quickdraw

Achievements

Achievement: YOLOAchievement: StarstruckAchievement: Quickdraw

Block or report pras-ops

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.

Content in all repositories owned by your account will be closed.
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
pras-ops/README.md

Hi 👋, I'm Prashant Jacob

Applied AI & Automation Engineer · LLM Agents · RAG · On-Device AI

I build applied-AI and automation systems that solve real problems — LLM agents, RAG pipelines, and privacy-first, on-device AI. Currently an AI Automation Engineer shipping GenAI workflows in production.

LinkedIn Portfolio Email

🚀 About Me

  • 🔭 Building applied AI & automation systems — LLM agents, RAG pipelines, and on-device / WebGPU AI
  • ⚙️ Currently an AI Automation Engineer working with n8n, MCP servers, and GenAI creative pipelines
  • 🌱 Exploring agentic architectures, local LLM inference, and privacy-first AI
  • 💬 Ask me about LLMs, RAG, AI agents, Python automation, and web scraping at scale
  • 🌐 Portfolio

🛠️ Tech Stack

AI / GenAI

LangChain LlamaIndex RAG MCP Hugging Face OpenAI Gemini scikit-learn

Automation & Web

n8n Playwright Selenium WebGPU / WebLLM BeautifulSoup Chrome Extensions

Languages

Python TypeScript JavaScript SQL

Backend & Data

FastAPI PostgreSQL SQLite Pandas NumPy

Cloud & DevOps

GCP Docker Git GitHub Actions

📌 Featured Projects

Project What it does Stack
ScrapeWizard ⭐ CLI that finds the repeating data on any page and saves a re-runnable, self-repairing scraper recipe, with an optional LLM mode. Picks the right list on 11 of 14 unseen sites with no AI key. Python · Playwright · LLMs
retrieval-reputation-layer Reputation layer for any retriever: turns verified downstream outcomes into a ranking signal (Beta counters, Thompson sampling, staleness decay). Evaluated on 974 MBPP coding tasks against a cross-encoder reranker; on PyPI. Python · sentence-transformers · FastAPI · LangChain / LlamaIndex
redactkit Local privacy layer for cloud LLMs: redacts PII before a prompt leaves the machine and restores it in the reply. Ships as an SDK, a local proxy and a browser extension. JavaScript · transformers.js · Tesseract OCR
WebLLM Document Reader Chat with your documents fully offline in Chrome. Falls back from Gemini Nano to WebGPU (Qwen 1.5B) to WASM (SmolLM 360M), with local RAG and page-level citations. TypeScript · WebLLM · WebGPU
AnchorHeal Drift early-warning and self-healing layer for Python scrapers: tracks per-selector confidence to flag scrapers that are about to break, and relocates moved elements locally without an LLM call. Python · Selenium · Playwright · BeautifulSoup
udemy-transcript-extractor Privacy-first Chrome extension that batch-extracts course transcripts in RAG-ready formats. TypeScript · Chrome MV3

Pinned Loading

  1. ScrapeWizard ScrapeWizard Public

    AI-powered CLI that analyzes websites and generates real Playwright web scrapers with built-in data quality checks

    Python 39 2

  2. udemy-transcript-extractor udemy-transcript-extractor Public

    A browser extension to automatically extract and batch-copy transcripts from Udemy videos. Perfect for students and learners who use note-taking apps like NotebookLM.

    TypeScript 24 9

  3. redactkit redactkit Public

    Local privacy layer for cloud LLMs: redacts PII before a prompt leaves your machine and restores it in the reply. SDK, local proxy and browser extension, all on-device.

    JavaScript 1

  4. retrieval-reputation-layer retrieval-reputation-layer Public

    Reputation layer for any retriever: turns verified downstream outcomes into a ranking signal (Beta counters, Thompson sampling, staleness decay). LangChain and LlamaIndex adapters.

    Python