Hi, I'm Muhammad Ishaq

Leading the technical vision and product development for an AI platform that lets businesses build, deploy and scale AI-powered applications with minimal engineering effort — enterprise-grade infrastructure for AI agents, workflow automation, RAG and intelligent business operations.

• Architected and built the AI Cortexo Platform (AIP), enabling organisations to launch AI agents, AI workflows and intelligent automations from one unified platform.
• Designed multi-agent orchestration systems supporting autonomous task execution and enterprise workflows.
• Built scalable RAG pipelines powering enterprise document intelligence.
• Developed production-ready AI infrastructure with Python, FastAPI, Docker, PostgreSQL, Redis and AWS.
• Integrated OpenAI, Anthropic and open-source LLMs into enterprise workflows with intelligent model routing and cost optimisation.
• Implemented authentication, usage tracking, API management and scalable cloud deployment pipelines.
• Led product architecture, engineering strategy and AI research from concept to production, working directly with clients to automate business processes.
• Built and launched Wink, a local-first AI desktop agent for Windows (Electron, TypeScript): it sees the screen with a local vision model via Ollama, then either guides the user click by click or operates the PC itself, with on-device Whisper speech-to-text and Kokoro voice, so screenshots never leave the machine.

Building FlashSync AI, a pre-launch AI career-agent startup: first the production website and waitlist (from 4 design concepts to a live product on Vercel in about one week), now the product itself, the FlashSync Career Vault AI career agent.

• Built the Career Vault AI career agent (Next.js 15, React 19, TypeScript): a zero-knowledge, local-first vault where users keep their profile, roles, wins and saved jobs, encrypted in the browser before anything is stored.
• Designed envelope encryption with the Web Crypto API only: PBKDF2-SHA256 (600k iterations) derives a key that wraps a random AES-256-GCM data key; each record is sealed with AAD bound to its id and type, so tampered or swapped rows are rejected.
• Shipped passphrase auth with no server accounts (unlock, 15-minute idle auto-lock, passphrase change by re-wrapping the key) and encrypted backup export/import across devices on Dexie.js/IndexedDB.
• Added an AI STAR impact rewriter using Claude on Amazon Bedrock (EU region) that sends only the one bullet the user picks, plus a Manifest V3 Chrome extension that captures job posts from schema.org, Greenhouse and Lever pages.
• Developed Sync, a Gemini RAG chatbot with streamed answers, embedding-based guardrails that keep it on-topic, English and Urdu replies, and voice input and read-aloud.
• Engineered a custom viral waitlist with no third-party tools: referral-based queue ranking, early-access tiers, welcome emails, reCAPTCHA v3 and database-backed rate limiting on serverless functions with Turso.
• Reached PageSpeed scores of 97–99 on mobile and 99–100 on desktop, cutting mobile LCP from up to 2.9 s to under 1.9 s with prerendered SEO pages; 0 WCAG 2.1 AA violations.
• Delivered the QA suite used for client acceptance: Playwright across 6 browsers and devices, visual regression, a load test at 200 concurrent users with 0 errors, and 53 unit tests.

• Built and deployed scalable AI and data systems, including NLP-powered intelligent chatbots and automated data pipelines on AWS.
• Designed cloud data architectures with AWS Glue and Bedrock, using Docker-based deployments for production.
• Developed and optimised ETL workflows to process large-scale enterprise data for analytics and AI use cases.
• Engineered knowledge graphs to model complex entity relationships, reducing AI retrieval latency by up to 85% while maintaining high accuracy.

• Led the AI/Data Science team driving development of scalable Generative AI and SaaS applications.
• Built and deployed RAG pipelines with LangChain and AWS Bedrock, achieving 99.9% uptime and sub-second retrieval latency.
• Designed inference optimisation strategies that cut operational costs by 25% while improving throughput.
• Mentored junior engineers and apprentices, establishing structured practices for GenAI SaaS products.

• Designed and deployed Generative AI solutions using state-of-the-art LLMs and AWS SageMaker for fine-tuning and inference.
• Built multilingual NLP systems and computer vision pipelines focused on automation, accuracy and scale.
• Developed AI chatbots, speech-to-text pipelines and intelligent document processing systems for enterprise workflows.
• Implemented end-to-end ML workflows — data processing, training, deployment and monitoring — on ECS and EKS.