I'm Atul Singh — Founder & CEO of Cognitiev 🚀 — and a hands-on builder in the Agentic AI space. I've bootstrapped Cognitiev from zero to a portfolio of production AI products — lean, resourceful, and without external funding — while shipping code every single day.
My obsession: autonomous, goal-driven AI agents that plan, reason, use tools, collaborate, and self-correct. I design multi-agent architectures that go beyond chat — systems that do work: orchestrating workflows, grounding themselves in enterprise data, and operating safely with humans in the loop.
Doctor Cognitiev, my flagship product, delivers early diagnosis and continuous home therapy for neuro-divergent children — an agentic AI companion recognized by medical professionals.
- 🔭 Currently building: Multi-agent orchestration platforms & agentic RAG pipelines at Cognitiev
- 🤖 Deep in: LangGraph, MCP, tool-calling agents, evals & agent observability
- 👯 Open to collaborate on: Agentic AI & GenAI applications
- 🤔 Looking for help with: Sales & Marketing
- 💬 Ask me about anything — except Maths 😅
- ⚡ Fun fact: I invest most of my free time researching Defence Forums
| Capability | What it means in production |
|---|---|
| 🧩 Multi-Agent Orchestration | Planner–executor–critic topologies, supervisor graphs, agent handoffs |
| 🛠️ Tool Use & Function Calling | Structured outputs, MCP servers, API/tool routing, sandboxed execution |
| 📚 Agentic RAG | Hybrid retrieval, query decomposition, self-querying, re-ranking, citation grounding |
| 🔁 Reflection & Self-Correction | Critic loops, guardrails, retry policies, output validation |
| 🧪 Evals & Observability | Trace-level debugging, LLM-as-judge, regression suites, cost/latency budgets |
| 🧑⚖️ Human-in-the-Loop | Approval gates, escalation paths, confidence-based routing |
| 🧠 Memory Architectures | Episodic + semantic memory, vector stores, session state, knowledge graphs |
Production systems I architect and code end-to-end — frontend, backend, and AI agent layers.
Real-time, telephony-grade conversational AI agents that handle live phone calls, IVR flows, voicemail detection, and outbound dialing — with sub-second latency.
| Layer | Stack |
|---|---|
| 🤖 Agent Runtime | Python · LiveKit Agents + MCP tool servers · OpenAI · Deepgram STT · Cartesia & Azure TTS · Silero VAD · DeepFilterNet3 server-side noise cancellation |
| ⚙️ Backend | Node.js · TypeScript · Express 5 · Prisma + PostgreSQL · Redis (ioredis) · LiveKit Server SDK · Twilio & SIP trunking · Stripe billing · S3/R2 storage · Prometheus metrics |
| 🎨 Frontend | React 19 · TypeScript · Vite · LiveKit Components · Redux Toolkit + TanStack Query · Tailwind CSS 4 · Radix UI · Framer Motion · Recharts · Socket.IO |
| 🔊 Voice Pipeline | Streaming STT → LLM → TTS with barge-in, browser + server noise suppression, call transcripts, sentiment analytics, RAG knowledge base |
| 📦 Infra | Docker · docker-compose · GitHub Actions · SIP/PSTN inbound-outbound runbooks |
Full-stack CRM with an agentic AI roadmap — lead pipelines, QR-based engagement, and analytics dashboards.
| Layer | Stack |
|---|---|
| 🏗️ App | Next.js 16 (App Router) · React 19 · TypeScript · Tailwind CSS 4 |
| 🗄️ Data | Prisma · PostgreSQL · Zod validation · JWT auth (jose) |
| 📈 Product | Recharts analytics · QR code workflows · Dockerized dev environment |
🩺 Doctor Cognitiev — Early Detection. Brighter Futures.
AI-powered developmental screening and voice-guided home therapy for children aged 0–6 — built on validated clinical instruments (M-CHAT-R, Vanderbilt), with Red-Amber-Green risk triage, multi-language caregiver guidance, and DPDP-compliant data governance. Recognized by medical professionals.
💡 "Agents are not the future of software — they're the present. I just build them before the docs exist."
⭐ From crypticatul — building Cognitiev, one commit at a time.