AI Delivery Manager • AI Program Architect • Founder, SkillWisor & Cognivexta
Bangalore, India • 15+ years in AI, data & delivery leadership
Who I am
I’m Naresh Matta, an AI Delivery Manager and program architect with over 15 years of hands-on experience building, managing, and scaling AI and data operations programs for some of the world’s most demanding technology companies.
I currently lead a large-scale enterprise AI account at Qualitest, managing AI programs for Google — one of the world’s largest AI organizations. My teams span data annotation, prompt engineering, agentic AI systems, and quality assurance for production LLM pipelines. Previously, at Concentrix, I led delivery operations for clients including Apple and OnePlus across multiple geographies.
Beyond my corporate role, I am the Founder and CEO of Cognivexta, an AI consulting company specializing in turning enterprise workflows into agentic AI systems for global clients. I also founded SkillWisor, where I publish hands-on guides to AI engineering and AI delivery.
Professional background
- Current: AI Delivery Manager, Qualitest India — managing enterprise AI programs for Google (Aug 2022–present)
- Previous: Senior Delivery Manager, Concentrix India — Apple and OnePlus accounts (Apr 2012–Aug 2022)
- Founder & CEO: Cognivexta — AI consulting and agentic workflow systems
- Founder: SkillWisor — hands-on AI engineering guides
- Builder: Hackathon projects deployed on GCP: Vyonix Studio (Gemini 3 Hackathon) and Shivy AI / KlassbookAI (Gemini Live Agent Challenge)
- Researcher: Implemented TinyLLM-AttenRes — a from-scratch implementation of the Moonshot AI Attention Residuals paper (arXiv:2603.15031)
Areas of deep expertise
- AI program management: leading cross-functional AI delivery programs, LLM evaluation pipelines and data operations at scale
- Agentic AI systems: designing multi-agent architectures using Claude, Gemini and open-source LLMs with tool-calling, memory and orchestration layers
- LLM engineering: prompt engineering, RAG systems, fine-tuning, semantic caching, token optimization and model evaluation
- AI research implementation: research-paper-to-code implementations including custom PyTorch LLM architectures (GQA, Flash Attention, RoPE, KV cache)
- Data annotation & quality: 15+ years designing annotation workflows, inter-annotator agreement frameworks and quality control systems for NLP and multimodal AI
- Delivery & leadership: managing teams of 20+, client relationships with Fortune 10 companies, OKR-driven program execution
What SkillWisor covers
SkillWisor is where I write up what I build and test. Articles are reviewed against our Editorial Policy: tested code, primary sources and a last-reviewed date.
- AI Engineering — LLM internals, RAG, agents, evaluation and from-scratch implementations
- AI News & Analysis — what shipped in AI and what it means for builders
- AI Delivery & Careers — running enterprise AI programs and building a career in AI
Let’s connect
I’m always open to conversations about AI, consulting opportunities or collaborations. Reach me through the Contact page or on LinkedIn.