Forward Deployed Engineer
AI summary of the role
Forward Deployed Engineer owning the technical path from customer evaluation to design win for Quadric's Chimera GPNPU, a novel unified architecture for neural network, DSP, and control code.
What you’ll do
- Convert 2–3 active customer evaluations into committed design wins by getting customer workloads running at target latency and accuracy on Chimera.
- Port customer workloads onto Chimera — architect, adapt, profile, and write custom kernels when required.
- Work on-site and remotely with customer engineering teams: debug failures, hit benchmark targets, and guide model-architecture tradeoffs.
- Codify learnings into reusable assets — SDK skills, porting playbooks, and agentic content — to accelerate subsequent engagements.
What you’ll bring
- Strong C++ and Python; has written performance-critical code for a constrained target (DSP, NPU, GPU, or embedded SoC).
- Systems-level performance intuition — memory bandwidth, data movement, cycle-level reasoning.
- High agency and independent problem-solving; methodical and empirical rather than theoretical.
- Comfortable operating without a playbook and in ambiguity — this role is still being defined.
Technologies
C++ · Python · Chimera GPNPU · quantization · custom kernels · compiler · Claude Code · SDK · NPU · embedded SoC
About Quadric
Licenses the Chimera GPNPU — soft processor IP that runs neural-network, DSP, and control code on one unified architecture for edge and automotive SoCs.
Series C · 50–100 people
Source and classification
Technical pre-sales · Evidence for this classification:
Quadric Chimera GPNPU, which runs everything from CNNs to LLMs to arbitrary C++ on one core, with no fallback processor. Unlike a FAANG solutions role, you'll work the entire stack: quantization, custom kernels, and compiler and your fingerprints will be on chips that ship in cars, cameras, and network appliances. At a Series C company, every design win has outsized impact. Our FDEs also build with frontier agentic tooling (Claude Code–based agentic workflows) — you won't just be supporting the product, you'll be inventing how edge-AI field engineering is done. What You'll Do Own the Technical Path to Design Win Convert 2–3 active customer evaluations into committed design wins by getting customer workloads running at target latency and accuracy on Chimera. Port customer workloads onto Chimera — architect, adapt, profile, and write custom kernels when required. Embed with Customer
More from the job description
Description About Quadric Quadric is redefining edge AI with the industry's first General Purpose Neural Processing Unit (GPNPU), enabling developers to run both neural network inference and conventional C++ code on a single programmable architecture. Our technology powers intelligent edge devices across automotive, industrial, robotics, and embedded systems. Founded by technologists from MIT and Carnegie Mellon, Quadric is a well-funded, Series C semiconductor IP company with marquee customers including Denso, Kyocera, and Hitachi, and a growing licensing business. As we scale our field engineering function, we're looking for Forward Deployed Engineers to take our novel Chimera GPNPU architecture from customer evaluation to committed design win. The Opportunity You'll own the technical path from customer evaluation to silicon design and win on a genuinely novel architecture, the Quadric Chimera GPNPU, which runs everything from CNNs to LLMs to arbitrary C++ on one core, with no fallback processor. Unlike a FAANG solutions role, you'll work the entire stack: quantization, custom kernels, and compiler and your fingerprints will be on chips that ship in cars, cameras, and network appliances. At a Series C company, every design win has outsized impact. Our FDEs also build with frontier agentic tooling (Claude Code–based agentic workflows) — you won't just be supporting the p [... source excerpt omitted ...] venting how edge-AI field engineering is done. What You'll Do Own the Technical Path to Design Win Convert 2–3 active customer evaluations into committed design wins by getting customer workloads running at target latency and accuracy on Chimera. Port customer workloads onto Chimera — architect, adapt, profile, and write custom kernels when required. Embed with Customer Engineering Teams Work on-site and remotely with customer engineering teams: debug failures, hit benchmark targets, and guide model-architecture tradeoffs. Drive evaluations forward end-to-end, from first technical conversation through design win. Build the Agentic FDE Toolkit Codify what you learn into r [... source excerpt omitted ...] s — SDK skills, porting playbooks, and agentic content — so each subsequent engagement moves faster than the last. Build applications and benchmark collateral that support future customer engagements. Close the Loop Internally Partner with engineering and product teams to relay customer learnings and shape the roadmap. Contribute skills and playbooks to Quadric's growing agentic FDE toolset. What Success Looks Like Within your first 6–12 months, you'll: Convert 2–3 active customer evaluations into committed design wins on the Chimera platform. Build deep fluency across the Chimera stack — quantization, custom kernels, and compiler tooling — well enough to unblock customer
Employer postings · Data from · Sources