AI Engineer
500 Skills, Zero Fine-Tuning
How LinkedIn scaled searchable tools and reusable agent skills without fine-tuning.
Ajay Prakash
Senior Staff Software Engineer at LinkedIn, building AI agents and the systems around them.
A personal space for my writing, talks, and open-source work. I am passionate about AI and AI agents.
Portfolio
Selected work
Talks
Conference sessions on context engineering and AI agents.
QCon AI Boston 2026
Context Engineering at LinkedIn
How we built an MCP-based organizational context layer for AI agents at LinkedIn.
Writing
Articles and engineering perspectives published on LinkedIn.
LinkedIn Engineering · 2026
Contextual agent playbooks and tools
How LinkedIn gives coding agents organizational context through MCP tools and reusable playbooks.
LinkedIn · March 2026
Designing Efficient CLI Tools for AI Agents
Practical principles for agent-friendly command-line tools: non-interactive commands, structured output, dry runs, and concise skills.
LinkedIn · May 2026
Software Engineering Is Evolving Into Harness Engineering
Why engineering is shifting toward designing the context, feedback loops, memory, orchestration, and quality gates around AI agents.
LinkedIn Engineering · 2025
Building collaborative prompt engineering playgrounds
A Jupyter Notebook workflow that brought engineers and domain experts into the same rapid feedback loop.
Open source
Projects I build and share.
TypeScript · AGPL-3.0
Stratos
A framework for agents to author, run, and improve reusable TypeScript workflows—combining agents, MCP tools, and decision models.