Kausthubh J Rao
  • Blog

© 2026 Kausthubh

Back home

All Projects

14 projects across AI, systems, and open source.

infer.c: C/CUDA LLM Inference Engine

2025

CCUDALLMsInferencingSystemsFlashAttention

C/CUDA inference engine with Paged KV Cache, Speculative Decoding, cBLAS/cuBLAS, Batching, FlashAttention, CUDA Graphs, and systems-level optimizations.

GithubKnow more
Fin Agent: AI Financial Assistant

2024

AgentsMCPTool CallingFastAPINext.jsOrchestration

FastAPI + MCP financial agent with autonomous orchestration and client-side tool calling with Next.js frontend, built for the Google Hackathon.

GithubKnow more
Gitea: Active Contributor

2025–2026

GoOpen SourceSecurityREST APIsRuntime Stability

Active contributor to a self-hosted Git platform with 55,000+ GitHub stars, working on security, API design, and runtime stability. (Yes, I fix bugs in software that runs your company's Git server.)

GithubKnow more
gRPC-Go: Open Source Contributor

2026

GogRPCHTTP/2NetworkingConcurrencyOpen Source

Refactored bidirectional stream state management in the HTTP/2 transport layer, eliminating redundancy across shared state transitions.

GithubKnow more
X-Cortex: Digital Automation Engine

2024

AIAutomationSystem DesignRakuten Conference

Presented X-Cortex, a digital automation engine, to 1,000+ professionals at the Rakuten Conference. Automating day-to-day digital interactions.

GithubKnow more
LLM Optimization Research

2024

PythonPyTorchLLM InferenceQuantizationModel Compression

Optimized LLMs for edge devices using dynamic pruning, quantization, and distributed inference by making models fit where they shouldn't.

Know more
Black - Python Code Formatter Contribution

2025

PythonOpen SourceCode FormatterPython 3.12CompilerAST

Contributed a fix to Black, the uncompromising Python code formatter, resolving formatting issues with Python 3.12's new generics syntax. The fix ensures proper handling of type parameters when function parameters contain trailing commas.

GithubKnow more
CUDA-LSP - High-Performance Language Server

2025

GoLSPLanguage Server ProtocolCUDACompilerIncremental Parsing

A blazing-fast Language Server Protocol (LSP) implementation in Go for CUDA development, featuring incremental diffing, intelligent hover documentation, and context-aware autocompletion built from scratch.

GithubKnow more
Real Incognito - AI Data Anonymizer

2024

PythonGANMachine LearningPytorch

A GAN-based data anonymization system that preserves data utility while removing sensitive information. Won 2nd place in AI domain at hackathon.

GithubKnow more
Freix - Logistics Platform

2024

SvelteTypeScriptTailwindCSSLogisticsPytorch

A logistics platform that simplifies international shipping tracking and management. Won 1st place in Logistics domain at Hackfest 2024.

GithubKnow more
Locked in Reality - Treasure Hunt Game

2024

SvelteTypeScriptTailwindCSSUI/UX

An immersive treasure hunt game with a Windows boot menu-inspired interface, created for Incridea college event.

GithubKnow more
Incridea Client - College Event Platform

2024

Next.jsTypeScriptTailwindCSSReact

Frontend development for Incridea, the largest college event platform, as part of the core technical team.

GithubKnow more
Student Hub

2023

Next.jsPostgresGoogle Maps API

A comprehensive platform for student resource management and collaboration.

GithubKnow more
Portfolio Minimal

2025

Next.jsReactTailwindCSSTypeScript

A minimal and modern portfolio website showcasing projects and skills.

GithubKnow more