Power Optimization Engineer, AI Hardware

Tesla Tesla · Auto · Austin, TX · Tesla AI

Senior Power Optimization Engineer for AI Hardware at Tesla, focusing on RTL-stage power analysis and optimization for next-generation inferencing chips. The role involves using EDA tools to reduce power consumption through techniques like clock-gating refinement and datapath rebalancing, influencing architectural decisions, and collaborating with various design teams to achieve system-level power reductions for AI accelerators.

What you'd actually do

  1. Drive RTL-stage power analysis and optimization using a combination of industry RTL power tools, internal metrics, and final PD power estimates, and turn findings into accepted RTL changes across the SoC
  2. Reduce switching activity and glitch power at RTL through clock-gating refinement, operand isolation, datapath rebalancing, and pipeline insertion
  3. Work closely with performance team to understand potential performance impact of proposed RTL changes and evaluate tradeoffs
  4. Work with subsystem owners to build representative activity vectors for AI inference workloads and use them to drive power analysis and signoff with PrimePower, PowerArtist, Joules, RedHawk-SC, or Voltus
  5. Collaborate with physical design on clock-tree power, multi-Vt mix, MBFF strategy, and EM and IR signoff feedback into RTL, and timing impact of proposed RTL power optimizations

Skills

Required

  • 5+ years ASIC/SoC RTL design experience on large multi‑billion‑gate SoCs with focus on power‑sensitive blocks and awareness of DFT impact on power
  • Production experience with RTL power optimization using tools such as Synopsys PowerArtist, Cadence Joules, Synopsys PrimePower, or Siemens PowerPro
  • Hands-on experience with switching-activity and glitch power analysis using SAIF or FSDB driven flows from RTL or post-synthesis simulation
  • Strong knowledge of low‑power techniques including clock gating, multi‑Vdd, multi‑Vt, MBFF merging, operand isolation, UPF power‑intent flows and power-gating strategies
  • Experience interfacing RTL design with synthesis, place and route, and signoff power flows including activity vector generation for representative workloads
  • Familiarity with AI accelerator workloads including matmul and attention activity patterns, sparsity-aware power, DRAM and SRAM power dominance
  • Experience driving RTL power reduction through multiple tape-out cycles with measured before-and-after power numbers
  • Extensive RTL development using SystemVerilog/Verilog and scripting (Python/Tcl/Perl)
  • Familiar with industry tools: PrimeTime, RedHawk‑SC or Voltus, and emulation‑driven activity capture
  • Degree in Electrical Engineering, or equivalent experience

What the JD emphasized

  • power reduction across the SoC
  • RTL-stage power analysis and optimization
  • power-saving opportunities
  • power reduction throughout the design cycle
  • power analysis and optimization
  • power reduction
  • power impact
  • power analysis and signoff
  • power budgets
  • power reduction through multiple tape-out cycles
  • power-sensitive blocks
  • power optimization
  • power analysis
  • low-power techniques
  • power-intent flows
  • power-gating strategies
  • power flows
  • power dominance
  • power numbers
  • power-aware coverage

Other signals

  • AI inference chips
  • train massive neural networks
  • Full Self-Driving (FSD)
  • humanoid robot, Optimus
  • custom silicon and optimized architectures