About cmul8
cmul8 puts agents to work inside the enterprise: an ontology of your data and the world's, agents adapted to your context, and the governance to run them in production — from simulations to decisions to shipped products. The company is based in Bengaluru.
The team
Core team, Bengaluru.
Founder and CEO of cmul8. Previously VP of Data & AI at Jetapult, building and shipping AI products across a global gaming portfolio. Founded MachineHack, Asia's largest community of AI professionals. Research stint at IISc Bengaluru. Also runs Coding Agents HQ, a practitioner community for agentic engineering.
Leads engineering across the platform — the engine, the agents, and the per-client apps they power. Founder of QuickCall, contextual memory for agentic engineering teams. Built India's first competitive Marathi LLM. Years of hands-on LLM-systems work, from data pipelines to deployed inference.
Leads research at cmul8. M.Tech (Research) in Computer Science & Automation at IISc Bengaluru. Owns the knowledge graph and the model calibration — the technical core of the platform. Published at top AI/ML venues.
Advisors
Founder and Chief Scientist of Fast Code AI, and Adjunct Professor of Computational and Data Sciences at IISc Bengaluru. PhD from the Max Planck Institute; postdoctoral research at NYU with Yann LeCun. Two decades across machine learning, computer vision, and graphics, with 8,000+ citations. Co-author of cmul8's multi-agent research.
Director of Technology Research & Discovery at Target. Two decades of enterprise technology leadership at retail scale. Advises cmul8 on enterprise strategy — how the platform earns its place inside large organisations, and what it takes to keep it there.
Capabilities
Ontology → Agents → Outcomes.
Ontology
The outside world catalogued and your internal data organised into one governed map — external signals, enterprise data, and simulated populations, kept current and queryable.
Agents
Built on the ontology, calibrated per domain, adapted to your context — and governed in production, with access control, audit, and escalation enforced by the engine underneath.
Outcomes
Decisions with provenance, simulations that rehearse choices before the real world grades them, and custom apps shipped per client — grounded in published research on multi-agent reliability.
Working on something where decisions need receipts? → Start a conversation