Cisco grants 27-year-old engineer unlimited AI tokens for 98% legacy code refactor in 24 months

Patel discovered Krishnamurthy, 27, five levels down after Cisco's 2025 acquisition of EzDubs. The former founder received four weeks to demonstrate feasibility followed by a 24-month target. Patel bypassed senior teams that viewed the scale as unrealistic and instead empowered recent hires to drive the effort.

No public benchmarks exist for Cisco's codebase volume or prior refactor metrics. Patel's approach aligns with patterns at firms like Google and Meta where AI-assisted migration reduced manual effort by 40-60% in controlled pilots. The decision prioritizes velocity over established review gates.

Operationally the move compresses maintenance cycles and reduces dependency on long-tenured staff for routine modernization. It also creates explicit tension between junior and senior cohorts to surface unfiltered questions. Success hinges on token consumption staying within internal cost thresholds and output passing security and reliability audits.

Cisco will expand recruitment of similar profiles from college talks and track monthly progress against the 24-month deadline. Parallel tracking of defect rates and deployment frequency will determine whether the model scales beyond the initial team.