Ayush kumar
Lead Agentic AI Architect | AI Security | AI Researcher | Building Autonomous Agentic SOC | Pioneered SuperAGI | Ex-Research and Development Intern @ Samsung | Ex-ML Algorithm Engineer
Greater Bengaluru Area
4+ years of work experience
Mar 20251 yr 6 mosPresent
Bengaluru, Karnataka, India · Hybrid
I lead how our organization builds, secures, and adopts AI agents at enterprise scale, turning frontier AI into production systems that dozens of teams and thousands of employees rely on every day.
• Architected and Built the company-wide standard for how AI agents securely connect to enterprise systems: a secure AI platform of 15+ connected services and 240+ tools (built on the Model Context Protocol) that every AI-to-system integration in the company now follows.
• Designed and shipped a five-tier multi-agent AI system that autonomously handles 800+ operational tasks a month across 25 services, cutting average resolution time by 88% (from 45 minutes to 4).
• Created the governance model for AI and non-human identities, defining how autonomous agents are trusted, permissioned, and contained so they can act at scale without becoming a security risk.
• Built an AI-native quality and analytics platform behind VP and Director reporting that cut manual quality work by around 95% and restored accurate, trustworthy compliance visibility.
• Established how every production AI agent is monitored, logged, and audited, with full decision replay so each autonomous action stays transparent and reviewable.
• Drove the organization-wide rollout of enterprise Security first generative AI Standards, from deployment and security baselines to hands-on enablement, giving the whole company governed access to AI.
• Recognized as the go-to person for agentic AI across security and engineering teams: I run knowledge and hands-on sessions, lead design reviews, and have helped 80+ engineers build with AI safely and confidently.
May 202410 mosFeb 2025
Bengaluru, Karnataka, India · Hybrid
Built the automation and platform foundations that let security teams ship faster and more safely.
• Led the organization-wide shift to Infrastructure-as-Code, automating how security platforms are configured and deployed and removing the conflicts that used to slow four teams down; presented to leadership and adopted company-wide.
• Redesigned and automated data onboarding for the security operations center, replacing manual, ticket-driven work with a self-service, approval-based workflow.
• Engineered 25+ complex data integrations and set the enterprise framework and quality bar that all later work followed, building the operational foundation for security analytics.
• Established the team's engineering-experience practices: reusable building blocks, safety guardrails, and standardized templates that cut repetitive work and lifted reliability.
Jul 202310 mosApr 2024
Bengaluru, Karnataka, India · Hybrid
Laid the data and visibility foundations for the security organization.
• Built the company's enterprise security data lake from the ground up, unifying 50M+ data points from across the business into a single source of truth for security analytics.
• Rebuilt the core security logging pipeline end-to-end, delivering 31 high-priority data onboardings for reliable, standardized data flow.
• Shipped real-time security dashboards that gave operations teams and leadership live visibility and faster response.
Apr 20234 mosJul 2023
Bengaluru, Karnataka, India
One of the earliest open-source frameworks for autonomous AI agents, built when the idea barely existed.
• Pioneered and Envisioned SuperAGI as a founding engineer. It grew to 21,000+ GitHub stars and was adopted by 15,000+ developers worldwide.
• Built the core agent orchestration engine, the reason-and-act loop that lets an agent take a goal, choose the right tool at each step, and work through multi-step tasks on its own.
• Designed the tool-integration system that connects agents to external apps, APIs, and third-party services, and shaped it into a marketplace of capabilities developers could plug into their own agents.
• Built the agent memory layer on top of multiple vector databases, so agents could store context, recall earlier steps, and improve with every run.
• Engineered the backend that provisions, runs, and scales many autonomous agents at once, reliable enough for both local experimentation and production use.
• Added performance monitoring and token-usage optimization, giving teams visibility into how agents behave and real control over cost.
• Worked across the whole backend in a fast build, test, and ship cycle, and helped grow an active open-source community through documentation, examples, and hands-on contributor support.
• Architected and shipped SuperCoder, an agentic coding tool used by 5,000+ developers. It was autonomous software development that predated and prefigured today's mainstream agentic coding tools like Claude Code, Codex, and Lovable.
May 20217 mosNov 2021
Bengaluru, Karnataka, India
Worked on Development of Algorithm for the MEC Server Placement for 5g network in edge computing
Developed the Algorithm and logic from scratch and implemented it
Jan 202110 mosOct 2021
I'm delighted to have been selected among HPAIR Indian delegate and excited to learn and connect with the leaders of tomorrow.
HPAIR is an internationally recognized student organization at Harvard College, offering a forum of exchange to facilitate discussion of the most important economic, political, and social issues relevant to the Asia-Pacific region.
Jul 202010 mosApr 2021
Bangalore Urban, Karnataka, India
Subsystem head of Attitude determination and control system of the Cubesat at BMSCE Satellite club i.e. BMSCE Upagraha
Ayush’s AI Stack
Education
B.E, Electronics and instrumentation engineering
B. M. S. College of Engineering
20194 yrs2023
Join Ayush and 10k+ other professionals on
Tal is the app where bosses meet engineers & builders directly, without middlemen
Download Tal Boss