Each skill is a few short lessons, then practice. You rate yourself at the start and again at the end.
Core to the job
Chapter 2
AI chip architecture
Show whether memory or computing power slows an AI task
What you'll learn
Dataflow reuse
Roofline limit
Memory hierarchy
Speed power area
Practice
Label the Bottleneck in a Simple NPU Pipeline15 min
Size the On-Chip SRAM for a MAC Array20 min
Chapter 2
Digital logic
Design a step-by-step circuit in code to match written requirements
What you'll learn
Combinational vs sequential logic
Datapath and controller
Setup and hold time
Clock domain crossing
Accidental latches
Practice
Write and Simulate a Verilog Traffic-Light State Machine30 min
Fix a Clock-Domain Crossing Bug in a Provided Verilog Module40 min
Chapter 4
Functional verification
Write tests for a small chip design, including extreme cases
What you'll learn
Testbench roles
Directed and random
Coverage plan
Coverage gap
Practice
Write Directed Test Cases for a 4-bit Adder15 min
Build a Coverage Plan for a Simple FIFO20 min
Chapter 4
Chip design tools
Turn chip code into a circuit and check its size
What you'll learn
Design flow
Timing constraints
Critical path
Unconstrained path
Practice
Simulate a 2-to-1 Multiplexer in ModelSim or EDA Playground15 min
Synthesise a Counter in Vivado and Read Its Resource Utilisation Report25 min
Chapter 6
Git
Save clear changes to a project's history and request review
What you'll learn
Change states
Short-lived branches
Merge conflict
Useful commit message
Pull request review
Practice
Make a Commit: Track a Firmware File Change15 min
5 more in the plan
Technical writingWrite a one-page work summary and a handover noteCh 6
What you'll learn
State your assumptions
Trade-offs on the page
Reproducible test report
Comment the why
App-note structure
Draft or record
Precise procedure step
Practice
Write a Handoff Email for the Next Engineer10 min
Problem solvingTest each possible cause of a problem instead of guessingCh 6
What you'll learn
Problem vs symptom
Clean problem splits
Root cause
Deciding under uncertainty
CommunicationRewrite an update for two different readers, cutting extra linesCh 7
What you'll learn
Lead with the point
Match detail to the reader
Said versus heard
Specific feedback
Practice
Explain a circuit concept to a non-engineer15 min
Working with othersContribute to group work, ask for help and credit teammatesCh 7
What you'll learn
Early team friction
Clear ownership
Challenge ideas, not people
Lead or follow
Practice
Write a Request That Makes It Easy to Say Yes10 min
Adapting to changeTry a new tool and describe how you handled changeCh 7
What you'll learn
Change vs your response
The messy in-between
Not there yet
Real change vs a feeling
Your plan
Chapter by chapter.
11 chapters, in this order. Chapter 1 is free.
1~2 wks
See the work of an AI silicon and accelerator design engineer
StartFree
2~2 wks
Design digital logic for AI chips
Skills
3~2 wks
Write a design review of your build
Project
4~2 wks
Build chip designs and prove them correct
Skills
51–2 wks
Talk to two working chip designers
People
6~2 wks
Write technical documents and track every version
Skills
7~2 wks
Communicate and work together through change
Skills
8~2 wks
Capstone: model an accelerator dataflow and compare designs
Project
91–2 wks
Answer the questions interviewers ask
Interview
101–2 wks
Decide how you will get in
Decide
11on your timeline
Send your first applications
Apply
The job
What a AI Silicon and Accelerator Design Engineer does
Designs the specialised chips that run AI models fast and efficiently, called neural processing units (NPUs) and AI accelerators. The work balances speed, power use, and chip area, then drives the design toward tape-out (sending the finished chip to be manufactured).
The work
Design the microarchitecture, the internal plan of an AI accelerator chip
Write RTL code, a low-level description of the chip's logic, that the hardware is built from
Plan dataflow so matrix math, the heavy maths in AI, runs efficiently
Balance speed, power, and chip area to hit performance targets
Tune memory bandwidth so data reaches the compute units fast enough
Run simulations and verification to catch bugs before silicon is built
Review designs with architecture, verification, and physical design teams
Document design choices and test results so others can reuse them
A typical day
Review performance targets, open bugs, and priorities with the team
Code or refine register-transfer level (RTL) logic for a compute block
Run simulations and check whether matrix math meets speed targets
Join an architecture review to weigh power, area, and speed tradeoffs
Debug a failing test using waveforms, logs, and performance counters
Write design notes and plan the next verification or tuning step
How long does it take to become a ai silicon and accelerator design engineer with Welica?
The plan is 11 chapters, 4–5 months at 15 to 30 minutes a day. You can go faster or slower; the plan moves with you.
Do I need a degree?
Not always. Common routes are bachelor degree in electrical engineering, computer engineering, or electronics, master degree focused on computer architecture or very large scale integration (VLSI) chip design, strong portfolio in RTL, verification, and computer architecture plus self-study, or graduate study in AI hardware, accelerators, or digital systems design.
What is free?
Chapter 1, See the work of an AI silicon and accelerator design engineer, is free for good. Premium unlocks the rest of the plan.
Start the AI Silicon and Accelerator Design Engineer plan.