Software engineer career path

You already split work into tickets, run design reviews and unblock people. Management makes that the whole job, then adds hiring, performance reviews and staffing calls.

Usually when

A year or two into a senior role, usually after you've run a project informally or onboarded a few new hires

Skills to add

• Structured 1:1s and written performance feedback
• Hiring loop design and interviewer calibration
• Quarterly planning and estimation across a team
• Incident command and postmortem facilitation

Have this on your resume first

One project you led with several engineers on it, what shipped, and who you mentored along the way.

If other teams already cite your design docs, you're doing staff work without the title. The rest is picking a technical direction and getting people you don't manage to follow it.

Usually when

Three or more years after senior, once teams you don't sit with are building on your designs

Skills to add

• Writing RFCs and design docs that get adopted
• Architecture decision records
• Planning multi-team migrations
• Capacity and cloud cost modeling

Have this on your resume first

A technical change you proposed and pushed through more than one team, with how far it spread and what it fixed, in numbers.

SRE is software engineering aimed at production. Your debugging and automation habits carry straight over, and SLOs (uptime targets), capacity and toil-killing tools are the new parts.

Usually when

Any point after 2-3 years, and sooner if you're the one who volunteers for on-call

Skills to add

• SLO and error budget design
• Prometheus and Grafana
• Kubernetes (Certified Kubernetes Administrator)
• Terraform
• Linux performance tools (perf, strace, eBPF basics)

Have this on your resume first

Incidents you resolved and reliability work you shipped, each with a latency, availability or recovery-time number.

Most ML engineering is plain production software, meaning pipelines, serving, tests and monitoring. If you already ship reliable services, you mainly need the modeling and evaluation layer.

Usually when

2-4 years in, ideally after you've built a feature that uses a model's output

Skills to add

• PyTorch
• scikit-learn and model evaluation metrics
• MLflow or Weights & Biases for experiment tracking
• Model serving with BentoML, TorchServe or a managed cloud endpoint

Have this on your resume first

A model you trained and deployed behind a real endpoint (an internal one counts), with latency and quality numbers.

Pushing back on scope, writing specs and talking to the people who call your API is already part of a technical PM's week. The trade is that you stop owning code and start owning whether the thing worked.

Usually when

4-6 years in, usually after you've become the engineer who argues about scope in roadmap meetings

Skills to add

• Customer discovery interviews
• Writing PRDs with success metrics
• Product analytics in Amplitude or Mixpanel
• Prioritization frameworks such as RICE

Have this on your resume first

A feature where you shaped the scope or the decision to build it, and the user or business metric that moved afterward.