// hello world

ANAGHA R S

Aspiring AI / ML Engineer

Contact →
scroll ↓

// 01 about

A little about me

I'm Anagha R S, an AI/ML enthusiast and open-source contributor focused on building intelligent, practical, and scalable solutions.

I enjoy transforming data into insights, experimenting with machine learning systems, and contributing to projects that create real-world impact. Currently, I'm deepening my expertise in AI, machine learning, and software engineering through continuous learning and hands-on development.

0 Projects Built
0 Certifications
0 OSS Contributions

// 02 experience

Where I've worked

Dec 2025 — May 2026

AI/ML Research Intern

Indian Institute of Technology Madras · Worked on building and fine-tuning ML models for industry use cases, collaborating with cross-functional teams.

// 03 education

Where I've studied

2024 — 2026

M.Tech in Computer Science & Engineering (Data Science & AI)

Cochin University of Science and Technology, Kerala

2020 — 2024

B.Tech in Computer Science & Engineering

Your School Name, Kerala

Until 2019

Secondary Education

Vimala Hridaya Higher Secondary School for Girls, Kollam, Kerala

// 04 skills

Technologies & Expertise

Python
Python 95%
Java
Java 80%
C
C 75%
SQL
SQL 78%
PyTorch
PyTorch 88%
TensorFlow
TensorFlow 82%
Hugging Face
Hugging Face 85%
LangChain
LangChain 83%
Qdrant
Qdrant 78%
RAG
RAG Systems 87%
LLM
LLMs 90%
FastAPI
FastAPI 85%
PostgreSQL
PostgreSQL 80%
API
REST APIs 88%
RTC
WebRTC 70%
AWS
AWS 75%
Docker
Docker 82%
Linux
Linux 80%
Git
Git 88%
GitHub
GitHub 88%
NumPy
NumPy 85%
Pandas
Pandas 85%
Scikit-Learn
Scikit-Learn 83%

// 05 projects

Featured Projects

01

OSS Copilot

AI-powered platform that helps developers discover open-source projects and contribution opportunities. Built with RAG pipelines, semantic search, and agentic workflows to analyze repositories, recommend issues, and generate contribution guidance.

LangGraph LangChain FastAPI PostgreSQL Qdrant
GitHub →
02

Predictive Human Activity Modeling

Developed a hybrid Graph Attention Network (GAT) and BiLSTM model for smart-home activity prediction. Achieved 98% activity accuracy and 0.93 macro F1 while forecasting time-to-next-activity across sensor streams.

PyTorch PyG GAT BiLSTM Time Series
GitHub →
03

Legal Document Summarizer

NLP-powered document summarization system using BART and Hugging Face Transformers. Processes lengthy legal PDFs in seconds and exposes a FastAPI endpoint supporting PDF, DOCX, and text-based inputs.

Python FastAPI Hugging Face NLP BART
GitHub →

// 05.5 open source

Open source contributions

12 Pull Requests
7 Repos Contributed
8 PRs Merged
2 Issues Reported
sharmavaibhav31 / arachnode
↗

Contributed to Arachnode's Aggregator Service by implementing centralized date normalization, ensuring consistent handling of timestamps across multiple sources.

PR Merged #PR-62 Multi-Format Timestamp Support ↗
⭐ 28 Python
1 Contributions
ionfwsrijan / PatchPilot
↗

Developed backend and machine learning enhancements for PatchPilot's security scanning and finding prioritization pipeline.

PR Merged #PR-30 Findings Retrieval & Verification APIs ↗
PR Merged #PR-44 Severity Ranker Training Pipeline ↗
PR Merged #PR-65 Startup Ranker Loading & Scan Scoring ↗
⭐ 12 Python FastAPI SQLite ML DevSecOps
3 contributions
AditthyaSS / iloveAgents
↗

Opened and implemented Issue #204: Added automatic timeout handling for stalled API requests, improving reliability across LLM providers.

Issue #issue-204 Stalled API Request Handling ↗
PR Merged #PR-226 Stalled API Request Handling ↗
⭐ 90 Python
1 contributions
hudazaan / kuberef
↗

Added support for scanning YAML manifests recursively across nested directory structures.

PR Merged #PR-20 Recursive Manifest Scanning ↗
⭐ 3 Python
1 contributions
Eshajha19 / Algo-Infinity-Verse
↗

Added an interactive Recursion Learning Page featuring concepts, visualizations, code examples, and practice exercises.

PR Merged #PR-255 Recursion Learning Page ↗
⭐ 9 HTML CSS JavaScript
1 contributions
imDarshanGK / localmind
↗

Added a session cloning API endpoint for duplicating chat sessions and preserving conversation history.

PR Merged #PR-380 Session Cloning Endpoint ↗
⭐ 5 Python FastAPI REST API Pytest
1 contributions
Th-Shivam / VYOM
↗

Opened and got assigned Issue #71 to build an LLM-based conversational memory summarization system using Groq.

Issue #issue-71 LLM-Based Conversational Memory Summarization ↗
⭐ 33 Python Groq API LLMs Conversational AI
1 contributions

// 06 certifications

Certified & verified

// 07 blogs

Thoughts & writings

Jun 2025

From Idea to Code: Making a Neural Network Actually Learn XOR (with Just NumPy)

Building a neural network from scratch using NumPy and understanding how backpropagation enables a model to learn the XOR problem.

Read more →
Jun 2025

One Model to Rule Them All? Not in Machine Learning 🤖

Why there is no universally best machine learning model and how choosing the right algorithm depends on the problem and data.

Read more →
May 2025

Neural Networks Aren't That Mysterious — They're Just Stacked Linear Regressions (Until They're Not)

Breaking down neural networks into intuitive building blocks and exploring how non-linearity transforms simple models into powerful learners.

Read more →
May 2025

My First Dive into Supervised Learning (with 10,000 Emails!)

Lessons learned while building a supervised learning project, inspired by The Hundred-Page Machine Learning Book by Andriy Burkov.

Read more →

// 08 contact

Let's connect

Have a project in mind, a question, or just want to say hi? My inbox is always open.