Samy
Kondasamy Jayaraman
Engineering Leader · Data Platforms & Agentic AI
I build the teams and platforms behind reliable data products and AI systems. Over 14 years, I've led engineering across data platforms, fraud systems, and agentic AI. Today, I architect operational intelligence at Mastercard.
Selected writing
View all →DeepSeek Harness, Cordis, and the Case for Spatiotemporal Agent Architectures
DeepSeek AI and Peking University formalized agent modularity in a recent paper on spatiotemporal composability. Here is an engineering look at the Cordis kernel, revertible effects, and what this pattern means for self-modifying agents.
Read the essay →
Today I Learned
View all →Publishing Activity
Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug
Mon Wed Fri
Blog: Building Voice AI Agents: Why Text Platforms Can't Just 'Add Voice', TIL: Time Travel Queries in Apache Iceberg, TIL: Rust Ownership Rules SimplifiedBlog: I Built a $500 SaaS Boilerplate in 5 Days (And Made It Free)TIL: How to Set Up Multiple SSH Keys for GitHub on MacBookBlog: Debugging a 500 Internal Server Error in Kubernetes: A Detective StoryTIL: Using pnpm outdated to Check and Upgrade DependenciesBlog: The Hard Parts of Voice AI: Latency, Prompts, and the Pink Elephant ProblemBlog: Beyond the Sci-Fi Brain: 6 Design Patterns That Power Real AI AgentsBlog: Claude Code: Things That Made Me Switch from CursorTIL: Git Worktree: The Secret Weapon for AI Coding Agents, TIL: Making Git Worktrees Painless for Parallel AgentsBlog: How Claude Code's Interview Mode Changed My WorkflowBlog: The Fish Shell Terminal Stack: From Slow to FastBlog: Stop Adding Redis: Fix Your Missing Postgres Indices FirstTIL: Custom Status Line for Claude CodeBlog: Browser extensions: Plasmo Has 12k Stars. I'm Still Recommending WXTBlog: 5 Underhyped Claude Code Tips That Actually Ship CodeBlog: The Rust Rewrite: Modern Developer Tools I Lean On in 2026Blog: UV: The Python Tooling That Makes pip Feel AncientBlog: RAG for Dummies: What Nobody Tells You About Building Search That WorksTIL: Intercom S3 Backups: Dealing with Empty FilesBlog: What Is an AI Agent? The Definition War That Reveals Our Industry's ConfusionBlog: Workflow vs Agent vs Multi-Agent: The Architecture Decision Nobody ExplainsBlog: How Claude Code Actually Works Under the HoodBlog: Multimodal RAG: You Don't Need CLIP (Until You Do)Blog: Dense vs Sparse Retrieval: Why Your Vector-Only Search Is Missing 25% of ResultsBlog: grep Is Not BM25: The Difference Between Pattern Matching and RetrievalTIL: Claude Code's Edit Tool Has a Tabs ProblemTIL: sips: macOS Hidden Image Swiss KnifeBlog: Pros Don't Vibe, They Control: What Developers Reveal About AI Agent UseBlog: GSD vs SpecKit, BMAD, and Compound Engineering: Picking the Right AI Coding FrameworkBlog: LiveKit vs Pipecat: One Feels Like Stripe, the Other Feels Like AirflowBlog: Why Compound Engineering Is My New Default for Claude CodeBlog: Escaping AI Slop: Comparing Design Copilots From Skills to Figma PluginsBlog: Anam vs Tavus vs Hedra vs Simli: Choosing an AI Avatar Platform That Actually Looks Human, TIL: Install Docker on Ubuntu (Hetzner Server Setup)Blog: 15 Patterns That Keep Production AI Agents From Burning Down ProdBlog: Claude Code's Source Got Leaked. Here's What Agent Builders Should Actually Learn From It.TIL: Portless: Named URLs for Local DevTIL: AlphaXiv: Research Papers Without the Token WasteTIL: Mole: One Command, 48GB FreedBlog: DESIGN.md: The Missing File Between AI Agents and Consistent UITIL: macOS Preview Does PDF EditingBlog: Harness Engineering: Stop Blaming the Model, Fix the EnvironmentBlog: The Unreasonable Effectiveness of HTML as AI OutputBlog: Claude Code in Large Codebases: What Anthropic Got Right and What They Left OutBlog: Stop Adding AI to Your Product. Start Deleting Everything It Replaces.TIL: Understand Anything: Codebase Knowledge GraphsBlog: MicroVMs: The Data Isolation Layer AI Agents Have Been MissingTIL: ccstatusline: The Status Line I Should Have Found SoonerTIL: iTerm2 Scrollback LinesBlog: Generative UI: The Missing Layer That Kept AI Apps Trapped in ChatboxesBlog: The GAN Pattern for Agents: Why Anthropic Stopped Letting Models Grade Their Own WorkBlog: Stop Prompting for Diagrams: Two Skills That Actually WorkTIL: Kami: Constraint-Based Document Design for AIBlog: Headroom: The Missing Layer Between Your Agent and Your WalletBlog: Ponytail: Teaching AI Agents to Write Less Code (and Why It Works), TIL: Git* URL-Hackable Repo ToolsBlog: Dissecting AI Agent Memory: A Practitioner's Postmortem on mem0Blog: Formal Verification: The Proof Oracles Your AI Agent Doesn't Know It NeedsBlog: Agentic vs Agentive: The Paper That Gets Agency Right and Accountability WrongBlog: Taste Skill: Teaching AI Agents to Stop Building the Same Purple Landing PageBlog: Tau: The Open-Source Agent You Read Like a TextbookBlog: Pi Agent: The Minimal Harness That Became My Multi-Tool GlueBlog: How omp Runs Subagents DifferentlyBlog: Kimi K3: What a 2.8T Open Model Changes for EngineersTIL: Pi's /tree Is Not Claude's /resumeBlog: Herdr and the Agent Multiplexer Gap: Why tmux Alone Isn't EnoughTIL: JSONL Is the Format AI Agents Were MissingBlog: The Deslop Stack Fallacy: Why Style Rules Hurt Thinking QualityBlog: Hermes Agent Postmortem: Why My Personal Agent Lives on a VPSBlog: DeepSeek Harness, Cordis, and the Case for Spatiotemporal Agent Architectures
Open source
GitHub profile →Contribution graph unavailable. View on GitHub.