Skip to content
View cherimedz's full-sized avatar

Block or report cherimedz

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
cherimedz/README.md

Typing SVG



pixel cat



LinkedIn


🌿   from the field to the model, the curiosity never changed   🌿



🌺   ✦   🌿   ✦   🌺   ✦   🌿   ✦   🌺


I started out studying life at its smallest, most intricate level — the living world of zoology. Now I decode a different kind of complexity: the patterns, reasoning, and emergent intelligence inside large language models and agentic AI systems.

I'm drawn to systems that feel almost alive. And to the craft of making them reliable.


AI-focused Software Engineer at 4Labs Technologies, building production AI that actually ships — Gemini-powered candidate scoring, real-time WebSocket pipelines, async email systems, and an AI-driven interview analysis platform in the works. Outside of work: autonomous agents, RAG systems built around regional Indian recipes, vision pipelines for medical prescriptions.

Finalist — Allianz Digital University Awards, Best Digital Innovation for Social Impact 2024 🏆


🌿   Former zoologist → AI engineer — the curiosity about complex systems never changed, only the subject matter

🌍   Trilingual · English · Hindi · Malayalam

📖   Avid reader, proud member of a university reading club

🌺   From Kerala, India

🐈‍⬛   Cat mom — my debugging process involves talking to them

✨   Currently building five things simultaneously (send help)


🌺   ✦   🌿   ✦   🌺   ✦   🌿   ✦   🌺

cats playing with feathers
🐾   Kiara · Coco · Princess · Ella   —   the current supervisory board   🐾

🌿   ✦   🌺   ✦   🌿   ✦   🌺   ✦   🌿


🌱   RMS Phase 3  ·  4Labs Technologies
AI interview & video-analysis platform — speech-to-text + LLM scoring
Django Next.js Gemini API


🌱   Stock Market AI
LangGraph agentic pipeline for autonomous market reasoning
LangGraph FastAPI MongoDB Docker

🌱   Novaire  ·  Ongoing
Multimodal prescription extraction — VLMs, OCR, drug interaction flagging
Streamlit HuggingFace pytesseract


🌱   Astravea
LLM + RAG disaster relief assistant — multilingual field report summaries
LLM RAG FastAPI


🌺   ✦   🌿   ✦   🌺   ✦   🌿   ✦   🌺




LangChain LangGraph Gemini API HuggingFace ChromaDB Sentence Transformers RAG Prompt Engineering Streamlit Celery WebSockets Pydantic Jinja2


🌿   ✦   🌺   ✦   🌿   ✦   🌺   ✦   🌿


🏢   Enterprise

Recruitment Management System  —  Production · 4Labs Technologies  ·  Multi-tenant AI recruitment platform · 200+ REST APIs · ~40% improvement in candidate relevance

AI-powered candidate scoring and JD question generation via Gemini API. Automated resume ingestion via Microsoft Graph, real-time WebSocket push via Redis, async email queueing with Celery and SMTP, AWS S3 document storage, and resume-to-branded-PDF pipeline.

~35–40% improvement in candidate relevance · ~50% improvement in recruitment efficiency

Django Next.js MongoDB Gemini API Celery Redis WebSockets AWS S3 Jinja2


🌿   Personal & Open Source

Velune  —  Autonomous multi-agent AI · Plan → Execute → Critique → Remember pipeline · self-evaluating output · long-term semantic memory  ·  LangChain · Qwen 2.5 72B · ChromaDB

A structured planner decomposes goals into 3–5 prioritized tasks. A LangChain ReAct executor runs Thought → Action → Observation loops with web search, Wikipedia, arXiv, math eval, and Python REPL tools. A Critic agent scores output quality 1–10 and triggers reruns on low scores. ChromaDB handles persistent semantic memory across sessions.

LangChain Qwen 2.5 72B ChromaDB Sentence Transformers DuckDuckGo Streamlit

Rasaveda  —  RAG recipe intelligence · 108+ global & Indian recipes · pantry-to-plate matching · substitution suggestions · conversational chatbot  ·  ChromaDB · FastAPI

Semantic ingredient-to-recipe matching using ChromaDB and Sentence Transformers. Features AI-driven recipe improvement analysis, ingredient gap analysis, a RAG chatbot, and a FastAPI REST layer. Five hand-crafted Indian regional UI themes.

ChromaDB Sentence Transformers Qwen 2.5 7B FastAPI Pydantic Streamlit

Caelune  —  AI study assistant · PDF / DOCX / images / handwriting → summaries, flashcards & quizzes · 12 subjects · 10 languages · 3 difficulty levels  ·  Llama 3.2 Vision · HuggingFace

Transforms multi-format notes into summaries, flashcards, quizzes, and glossaries. Vision OCR via Llama 3.2 Vision for handwritten scans. Streaming responses, session history, and adjustable difficulty.

Qwen 2.5 7B Llama 3.2 Vision HuggingFace Inference API pdfplumber Streamlit

Halcyon  —  ATS-optimised resume & cover letter generator · JD tailoring · Resume + Cover Letter bundle · 5 export themes (PDF & DOCX)  ·  HuggingFace · ReportLab

Generates resumes, cover letters, meeting notes, and business reports from free-form text or file uploads. Section-by-section AI rewriting, job description tailoring, and five visual themes.

HuggingFace Inference API Zephyr / Phi-3 / Mistral ReportLab python-docx Streamlit

AI Stock Market Analysis Platform  —  Ongoing · Agentic market pattern recognition & signal generation · async task pipeline · containerised  ·  LangGraph · FastAPI · MongoDB · Docker

Production-grade agentic platform for autonomous reasoning across market data. OpenAPI-spec-first FastAPI architecture with Celery and Redis for async task processing, containerized with Docker.

LangGraph FastAPI MongoDB Celery Redis Docker

Novaire  —  Ongoing · Multimodal prescription extraction · structured data from prescription images · automated drug interaction flagging  ·  VLM · OCR · NER

Pipeline extracting structured medication data from prescription images using Vision-Language Models, OCR, and named entity recognition, with risk flagging for drug interactions.

Vision-Language Models OCR NER Streamlit HuggingFace

Seeker  —  🏆 Allianz Digital University Award Finalist 2024 · AI freelancing platform · intelligent skill matching & recommendation engine

AI-driven freelancing platform built for the Allianz Digital University Awards. Finalist — Best Digital Innovation for Social Impact 2024.


🌺   ✦   🌿   ✦   🌺   ✦   🌿   ✦   🌺


🏅  Google Cloud Generative AI Track · 2025

🏅  Google Cloud Data Analytics Track · 2024

🏅  IBM AI Developer Professional Certificate · 2021

🎓  MSc Computer Science — Data Analytics
      Kerala University of Digital Sciences · 2023–2025

🎓  BSc Zoology
      University of Kerala · 2020–2023


🌿   ✦   🌺   ✦   🌿   ✦   🌺   ✦   🌿

Let's Connect


contribution snake


🕯️   This page was built in the glow of a laptop screen, with a cat on the keyboard and a dog at my feet.   🕯️


🐈‍⬛   Kiara, Coco, Princess & Ella — the current QA team, always on duty (especially during Zoom calls)


🕯️   Forever loved, never forgotten   🕯️

🐾   Poochu · Kathu · Tom · Sebastien · Simba · Kitty · Nymeria · Prince · Aaliyah
🐾   Shinto · Bruno · Puppy

— my first code reviewers, loudest critics, and truest fans.


You were here. You mattered. I carry you in every line I write, in every late-night debug session, in every quiet moment between commits.

You taught me patience when things broke.
And you taught me that even in silence, presence speaks volumes.

I may not change the world in big ways, but because of you, I live in it with a fuller heart. You will all be part of my family. You taught me what love really means — and that's the foundation of everything I build.


twelve cats · three dogs — a lifetime of love

Built with rose-tinted glasses  ·  Powered by too much chai  ·  Supervised by cats

And always, always — in your honor. 🐾


Popular repositories Loading

  1. Polynomial-Regression-for-Salary-Data Polynomial-Regression-for-Salary-Data Public

    In this scenario, I have used polynomial regression to predict the appropriate salary for a Human Resource company that seeks to determine the salary for a new job position based on available data.

    Jupyter Notebook

  2. Logistic-Regression-Using-Campus-Interviews Logistic-Regression-Using-Campus-Interviews Public

    In this scenario, I have used Logistic Regression to evaluate students based on their test scores from the company's assessment, their percentage of marks, and interview performance and predict whe…

    Jupyter Notebook

  3. Data-Science-Introduction Data-Science-Introduction Public

    This repository contains a JupyterLite notebook designed to provide a comprehensive overview of various data science tools and libraries. This notebook serves as an educational resource, showcasing…

    Jupyter Notebook

  4. Stock-and-Revenue-Analysis Stock-and-Revenue-Analysis Public

    This project focuses on extracting and visualizing key financial data for Tesla and GameStop, two prominent companies in the stock market, by utilizing Python libraries such as yfinance, BeautifulS…

    Jupyter Notebook

  5. Kiara-s-Haven-Website Kiara-s-Haven-Website Public

    In my first web project, I used HTML, CSS, JavaScript, Java, and Bootstrap to create a dynamic site with robust backend functionality and responsive design, marking my first website created.

    HTML

  6. Automatic-Classification-of-Medication-Data-from-a-Sentence Automatic-Classification-of-Medication-Data-from-a-Sentence Public

    Forked from annapurna1702/Automatic-Classification-of-Medication-Data-from-a-Sentence

    This project addresses the automatic classification of medication data from text using NLP techniques. By combining Word2Vec embeddings with Naive Bayes classification, it aims to accurately identi…

    Jupyter Notebook