📍 India | Senior Software Engineer at Okta | 🛠️ Former Ona/Gitpod engineer (acquired by OpenAI) | 🔐 OpenFGA Core Maintainer | 🎤 International Speaker | siddhantkhare.com
Building infrastructure for AI agents - context efficiency, least-privilege security, production-grade tooling.
I wrote a book: The Agentic Engineering Guide - 10 parts, 33 chapters on the engineering decisions behind AI agent infrastructure. Context engineering, authorization, cost control, evaluation, adoption playbooks.
Free to read online. PDF & EPUB on Gumroad. Source on GitHub.
Peer-reviewed conference paper · Springer · ICDEC 2025 proceedings · Published August 2026
My first published research paper, based on LLMTraceFX, the open-source GPU inference profiler I built.
I developed the concept and methodology, conducted and validated the study, and wrote the original draft.
Authors: Shubhanshu Kushwaha, Mamata Samal, Siddhant Khare.
- 🧬 Distill - Deterministic context deduplication for LLMs. Clean context in ~12ms, zero LLM calls.
- 🔍 agent-trace -
stracefor AI agents. Capture and replay every tool call, prompt, and response. OTLP export to Datadog, Honeycomb, New Relic, Splunk. - 🧪 ContextLab - Open-source LLM context engineering toolkit: analyze, compress, visualize.
- 💾 TokenVM - High-performance runtime treating LLM KV cache as virtual memory with page-based eviction.
- 📊 KV-Cache Profiler - Profile LLM GPU memory needs before deployment, not after.
- 🔬 LLMTraceFX - GPU-level LLM inference profiler with kernel timing and AI-powered bottleneck detection.
- 🤖 Agentflow - Kubernetes for AI agents - orchestration runtime, prompt ops, security layer, observability, and cost-aware scheduling.
- 🌐 AI Agent Orchestrator - Multi-agent AI system on Cloudflare Workers + Containers.
- 🚨 OnCall Agent - AI-powered on-call triage agent. Pulls alerts, checks deploys, maps dependencies, reads runbooks, and ranks hypotheses by confidence. Built on OpenAI Agents SDK.
- 🔢 LLM Parallelism Explorer - Research tool for optimizing parallelism strategies in LLMs (MoE focus).
- 💰 CloudArb - GPU arbitrage platform for AI compute optimization.
- 🔐 agentic-authz - OpenFGA + MCP authorization gateway for AI agents. Fine-grained access control at team, project, and tool levels.
- 🔏 Agentic Authorization - Authorization patterns for autonomous AI agent systems using ReBAC with OpenFGA.
- 🔗 A2AS Implementation - Proof-of-concept of Agent-to-Agent Security framework.
- 🛡️ Sentinel AI - Hermetic CLI for security scanning and dead-code detection with LLM-powered triage.
- 🔒 actionsec - Fast, local-first CLI for GitHub Actions security analysis.
- 🏗️ Cloud Architect AI - AI-powered cloud infrastructure design with architecture diagrams and Terraform code.
- 📝 Apple Notes MCP Server - Interact with Apple Notes via natural language.
- 📸 Memory Journal MCP Server - Search and analyze your photo library with AI.
- 🐳 Devcontainer MCP Server - Manage DevContainers using AI prompts.
- 📰 Dev.to MCP Server - MCP server for Dev.to.
- 🔧 Contextify - Lightweight script for LLM context injection from project files.
- 🧹 Gitpod Environment Cleanup - GitHub Action to find & delete stale Gitpod environments.
- 📈 Gitpod Runner Analyzer - Gitpod Flex runner resource analyzer.
- 🐙 GitHub Actions Self-Hosted Runner - Debugging GitHub Actions using Gitpod.
- 🏷️ Get GitHub Actions SHA - Fetch commit SHA for GitHub Actions tags.
- 🕵️ GitHub Venture Scout - AI-powered GitHub profile analyzer for identifying investment-worthy projects.
- 🇮🇳 Sponsorable GitHub Users in India - Top 1,000 sponsorable GitHub developers in India, auto-updated daily.
- ⚡ Scalable LS - List directories with millions of files (Rust).
- 🖼️ Image Credential Masker - Identifying & obscuring personal/sensitive information in images.
- 💳 Stripe Overdue Invoice Cleaner - Automatically void overdue Stripe invoices.
- 🌱 Greenhouse Data Exporter - Export candidate data from Greenhouse.io to CSV.
- 📜 Patent Summarizer - Patent summarization using OpenAI.
- 🍔 Swiggy Party Agent - AI agent that orders food for your entire team.
- 👁️ Gemma Vision - Local image intelligence CLI powered by gemma4:12b via Ollama.
- 🎬 Lipsync Check - Detects audio-visual desync in video files using Gemma 4 via Ollama.
- 🎵 Song Vector Explorer - Explore song lyrics as interactive 3D vector spaces.
- 🗺️ SageMap - Interactive tool to map and evolve personal beliefs.
- 🏥 Radiology Copilot - Gemini-powered multimodal radiology assistant.
- 🏛️ ArchiFusion - Transforms architectural ideas into 3D building models.
- 💊 MediSearchAI - A smarter way to search for medicines.
- 📄 MedBrief - Automated PubMed research paper summarization into narrated videos.
- 📚 LangChain x OpenAI: Bring Your Own Data - Train with custom markdown data using LangChain.
- 🖥️ GPT-CLI - GPT in your terminal.
- 🔑 OpenFGA - Core Maintainer of Google Zanzibar-style fine-grained authorization system (CNCF Incubating). First independent maintainer.
- 🔍 GitHub1s - Maintainer of one-second code reading for GitHub repositories.
- 📍 Geo-Location Attendance System - Geo-based attendance system for college students (Dart/Flutter).
- 🐬 MySQL Replica Server - MySQL master-slave replication setup in Docker.
- 💬 React Chat Application - Real-time online chat application with rooms.
- 📉 GitHub User Analytics - GitHub user analysis with charts using React, Auth0, and FusionCharts.
- 💹 Visualize Stock Market - Stock market data visualization.
- 🧫 Logistic Regression - Breast Cancer - ML classification on breast cancer data.
- 🛒 Amazon CLI - Amazon product search from the terminal using scraping.
- 📋 Sheets to CSV Converter - xlsx to CSV CLI converter (Go).
- 🗄️ S3 Backup Tool - Scheduled backups with cron syntax (Go).
- ⚙️ Slice vs Iterator Benchmarking - Slices vs Iterators benchmarking in Go v1.23.
- 📦 JSON to Table - JSON data to HTML table converter.
- 🔄 CRUD TypeScript - CRUD REST API using TypeScript, PostgreSQL, and TypeORM.
- 💱 Currency Converter - Currency converter CLI with MetaCall.
- 🎯 S3 Vectors Benchmark - AWS S3 Vectors performance benchmarking at scale.
- 🖌️ Satori Image Generator - Generate images from HTML/CSS using Satori.
- 📐 GitHub Contribution Tracker - CLI to fetch and format GitHub contributions to any org.
- 🔀 Sorting Algorithms Visualizer - Visual sorting algorithm comparisons.
- 🔌 Logic Gates Simulator - Interactive logic gates simulator.
- 🧭 Pathfinding Algorithms Visualizer - Pathfinding algorithm visualization.
- 🎨 Mandala Maker - Online mandala drawing tool.
- 🧩 JavaScript Sudoku Puzzle - Sudoku puzzle generator & solver.
- ♟️ Chess - Chess in HTML/CSS/JS.
- 🌍 Coronavirus Probability Checker - COVID-19 probability checker based on symptom data.
- 🩻 COVID-19 Rapid Tester - COVID-19 detector using chest X-ray analysis.
- 🔔 COVID-19 Outbreak Notification - Live notification alerts from official sources.
- 🚀 404 Error Page - Astronaut - Animated 404 error page.
- Engineering @ Ona - Building infrastructure for AI agents (formerly Gitpod).
- Open Source - Maintaining OpenFGA, GitHub1s, and shipping agent infrastructure tools.
- Writing - The Agentic Engineering Guide (book) + 40+ technical articles on AI infrastructure, security, and context engineering on siddhantkhare.com/writing.
- Speaking - KubeCon India 2025: Beyond Productivity: Scaling Cloud Dev Environments for Faster Feedback & Sustainable Engineering. Available for conferences on AI agent security and authorization.
- Mentoring - Book a session on MentorCruise.
- A cache hit is not proof that you skipped the work
- The compiled run had a lower request-latency sum. I still could not claim break-even.
- A benchmark result without lineage is just a screenshot
- The fastest passing system was not the cheapest one
- Parsing 7,585 XML references to count 400 Metal dispatches
- The weights fit. The inference workload didn't.
- A Metal trace is not your workload until you attribute it by process
- Your LLM's time to first token might be measuring your HTTP client
- A trace is not evidence if you don't know what it missed
- Building an Internal Platform for AI Agents
- OpenFGA Core Maintainer - First independent maintainer of a CNCF Incubating authorization project
- KubeCon India 2025 Speaker - Beyond Productivity: Scaling Cloud Dev Environments
- Former GitHub Intern
- Maintainer - OpenFGA, GitHub1s, Greenstand, Indian Open Source Foundation
- The Agentic Engineering Guide - 33-chapter open source book on AI agent infrastructure (read online / PDF & EPUB)
- 40+ technical articles - Published on siddhantkhare.com/writing, multiple posts with 100+ reactions
- Interviewed by New York Times (Cade Metz) and NPR Here & Now on AI fatigue
- Business Insider - Featured essay + video segment
- Futurism - "AI Is a Burnout Machine"
- Techmeme - Front page feature
- Hacker News - #1, 450+ points, 310+ comments
- The Chosun Ilbo (South Korea's largest newspaper) - "AI Fatigue Grows Despite Productivity Gains"
- NDTV Profit - Video segment on AI burnout
- Computing UK - Quoted extensively on AI fatigue
- Syndicated across Yahoo, AOL, Business Insider Nordic/Africa/Italy, NewsBreak, daily.dev, Lobsters, and others
- Discussed on Precursor Ventures podcast by Charles Hudson and Mia Farnham
- Beyond Productivity: Scaling Cloud Dev Environments - KubeCon India 2025
- AI Agents Are a Security Nightmare (Here's the Fix) - Agent authorization deep dive
- Building a Medical AI Agent with Gemini 3 Pro - End-to-end agent build
Building at the intersection of LLM efficiency and agent security:
- Context Efficiency & Reliability - Deterministic algorithms for context deduplication and optimization (Distill, ContextLab, TokenVM)
- Agent Authorization & Audit Trails - Google Zanzibar-style authorization for agent-tool interactions (agentic-authz, OpenFGA)
- Adversarial Robustness & Observability - Detecting and mitigating attacks on agent tool-use pipelines (Sentinel AI, agent-trace, LLMTraceFX)