Revisited Mentorship Paths for Engineers in the AI Era
“Most of us aren't lost — we just need new maps.”
The past year has pushed engineers into a strange duality: half-excited by the power of LLMs, half-anxious about what’s left to own. I've mentored engineers going through this exact shift — some thriving, others frozen. And it’s clear:
We need new pathways. Not based on titles. Not based on seniority. But on how you want to grow alongside AI.
So I revisited my original mentorship framework and updated it for this new landscape as it seems there is no turning back to the normative / regular way of practicing software engineering.
This isn’t a ladder. It’s a loop model — learn, simulate, reflect, reposition. Choose the orbit that fits your context.
🧠 1. The AI-Augmented Builder
For those who love moving fast and building the future with tools in motion.
You’ll spend time with:
- 🧠 Prompt Engineering, Function Calling (OpenAI, Anthropic)
- ⚙️ LangChain, LlamaIndex, WebLLM
- 🖥️ In-browser agents, multi-modal experiences (voice, vision, code)
- 🚀 Deno, Hono, Next.js, local-first AI apps, in-browser AI.
Of course, this will include the guidance on choosing the best of both worlds for your next side hussle under my target-to-the-goal planning and navigating.
Your loop:
Prototype → Simulate → Refactor → Share → Repeat
🔍 2. The Systems-Oriented Curator
For those who find beauty in evaluation, architecture, and trust.
You care how AI systems behave in the real world. You debug not just the output — but the alignment, latency, and failure states.
Core Idea: Become a steward of AI systems — where safety, observability, and evaluation are your north stars.
You’ll go deep on:
- 🛠️ LLM evaluation tools: Trulens, Ragas, Giskard
- 🔍 Observability stacks: Langfuse, PromptLayer, OpenTelemetry for AI
- 🧪 Synthetic testing, simulation frameworks, persona agents
- 🧱 Infra-first systems: ScyllaDB, NATS JetStream, WASM modules in Elixir/Go or maybe OCaml w/ MetaProgramming (why not sci-fi, right?)
Your loop:
Observe → Instrument → Evaluate → Improve → Scale
🎯 3. The Impact-Driven Explorer
For those who want to build for people, not just for infrastructure.
You see AI as a tool for shaping education, journalism, governance, and art. You want to bring your full mind — not just your stack — into your work.
Core Idea: Reframe engineering as applied thinking across domains — with AI as a lens, not the spotlight.
You’ll explore:
- 🧭 Human-AI collaboration: explainability, inclusive design, narrative UI
- 🧩 Agentic systems: CrewAI, AutoGen, LangGraph
- 📚 Domain-specific RAG: policy, health, civic tech, accessibility
- 🗺️ Storytelling + analytics: Observable, DuckDB, Streamlit, HuggingFace Spaces
Your loop:
Immerse → Reframe → Co-create → Share → Lead
You’re not locked in. You can orbit all three paths depending on your phase, team, or purpose.
But when you feel stuck, ask:
- Do I want to build faster? → AI-Augmented Builder
- Do I want to stabilize what others build? → Systems-Oriented Curator
- Do I want to build where it matters most? → Impact-Driven Explorer
Each one is valid. Each one demands loop-based growth, not linear progress.
I’ll be sharing deeper notes, templates, and examples for each path. If you’re designing your next move — or mentoring others — this model is yours to remix.
→ For the preliminary step ahead about what's coming with my renewed mentorship programme, jump here. You can also revisit the original post on choosing your engineering journey.
Stay looped in.
– Baris