AI-assisted, human-owned
Dedicated agents help us plan, code, review, test, secure, document, and publish. We understand, verify, approve, and remain accountable.
14 Sep 2026 → 13 Sep 2027
365 Days to
FullStack AI Engineer.
Mobile · Full-Stack · Applied AI
Our evidence-driven transition from React Native engineering toward Senior/Staff Mobile Engineer with full-stack and applied-AI capabilities—using TypeScript, Node.js, Python, Azure, and human-governed agents.
Experienced mobile and frontend engineers can follow with us. Freshers may use the same path, but should expect a longer foundation phase and different initial outcomes.
0
Calendar days
0
Weekly checkpoints
0
Planned milestones
0h
Weekdays + 75 min weekends
Our working product · reviewed in Week 1
Subra AI — a local Ollama agent
Our first working product is a small JavaScript agent that runs against a local Ollama model through its OpenAI-compatible API. It receives a goal, follows explicit instructions, can choose one allowlisted local tool, and reports the result for human review.
First usable slice: We can run the agent locally, stream its response, observe its model and token behaviour, and verify one bounded tool call without sending prompts to a cloud model.
Week 1 is intentionally small: one local model, one agent loop, and one safe tool. System prompts guide behaviour but are not treated as a security boundary; application code validates tool names, arguments, limits, and approval requirements.
Core technology map
One product. Two backend strengths.
React Native remains our product advantage. TypeScript and Node.js own the application backend; Python and FastAPI handle specialised AI workloads; Azure provides secure production operations.
The operating system
Our transition operating system.
We use dedicated agents for meaningful work from Day 1, while retaining the understanding, approval, and accountability required of a senior engineer.
Evidence over claims
Every phase produces working software, tests, architecture decisions, measured results, and an honest record of failures and corrections.
Build locally, operate on Azure
We develop locally with available tools, test local models where practical, and use bounded Azure deployments to learn identity, delivery, and operations.
Learn visibly from Day 1
We publish decisions and evidence throughout the year, compare our work with real openings, and improve how recruiters can evaluate it.
52-week roadmap
Assist. Understand. Ship. Prove.
We build on mobile from Week 3 and connect the backend in Week 5. Thursdays include repair and rotating coding, design, review, or career practice. Every fourth week we reassess readiness. Senior roles are the primary target; Staff scope requires demonstrated influence and collaboration beyond a solo project.
Local AI agent foundation
Weeks 1–4 · Workspace, one assistant, product brief, local-model trial, first mobile screen, and evidence workflow.
TypeScript backend systems
Weeks 5–11 · Mobile-to-Node delivery, PostgreSQL, ownership, one background job, containers, and consolidation. GraphQL is optional.
Python AI services
Weeks 12–17 · A small Python utility, stateless FastAPI service, summary evaluation, and human-reviewed mobile drafts.
Azure production foundations
Weeks 18–23 · Budget, identity, a small deployment, one trace and alert, infrastructure, and recovery.
Reliable LLM applications
Weeks 24–31 · Model evaluation, searchable entries, citations, access controls, and a measured RAG milestone.
Agents and MCP
Weeks 32–38 · Draft workflows, bounded tools, local MCP, approval state, one preference, and agent evaluation.
Mobile product hardening
Weeks 39–47 · Streaming, optional on-device and attachment experiments, architecture review, performance, and recovery.
Career evidence and applications
Weeks 48–52 · Milestone review, verified case studies, focused interview mocks, and an application cycle.
Portfolio evidence
Four planned milestones.
Each milestone extends the same user workflow. We verify acceptance checks and label local, test-deployed, and production-operated evidence separately.
First mobile screen and evidence workflow
A local journal screen, reviewed development loop, and evidence draft. This first milestone is a working foundation, with manual publication available.
Product brief → mobile entry → review → evidence draft
View build planConnected mobile journal and AI drafts
Mobile entries persist through Node and PostgreSQL. A bounded Python service prepares a summary that we can inspect, accept, or discard.
React Native → Node.js → PostgreSQL → Python/FastAPI → tests
View build planEvaluated RAG product
A cited-answer feature with access checks, failure analysis, and measured quality. We try at most one retrieval improvement when justified; Azure deployment remains pending if deferred.
Local models → Azure AI → retrieval → evaluation → observability
View build planReviewed mobile and agent product
The same companion gains bounded tools, approval controls, and operational evidence. We state which paths ran locally, were deployed for testing, or were actually operated in production.
Agents → MCP → mobile product → distributed systems → Azure operations
View build planDefinition of progress
The proof will be public.
From Day 1, weekly notes will separate what agents produced from what we understood and approved. They will show decisions, alternatives, tests, evaluations, failures, corrections, source code, demonstrations, and recruiter-facing case studies.
Common questions
Direct answers about the journey.
The essential facts for anyone deciding whether to follow or attempt the roadmap.
What is the 365 Days to FullStack AI Engineer journey?
It is our public, evidence-driven transition from React Native engineering toward Senior/Staff Mobile Engineer with full-stack and applied-AI capabilities. The compensation figure is a target, not a guarantee; the concrete outcome is stronger, reviewable engineering evidence.
Can freshers follow this roadmap?
Yes, with additional foundation time. The calendar assumes JavaScript/TypeScript and mobile or frontend experience. We repeat prerequisite work when needed; this plan does not establish a salary expectation for freshers.
Does the roadmap guarantee a particular salary?
No. Compensation depends on demonstrated skills, prior experience, location, the hiring company, role scope, and interview performance. The opening snapshots are market evidence, not a promise of an offer or salary.
What will be learned during the 365 days?
We will combine React Native and TypeScript with Node.js, PostgreSQL, GraphQL, Python/FastAPI AI services, Azure, RAG, agents, MCP, evaluation, security, observability, distributed-system design, and production operations.
How much time does the journey require?
We plan eight weekday hours: 90 minutes Monday–Thursday and two hours Friday. Saturday adds 45 minutes for evidence and Sunday adds 30 minutes for reflection: nine hours and 15 minutes total. Reading, agent interaction, checks, and notes are included. Unfinished core work changes future scope rather than expanding weekends.
What portfolio projects will be built?
We plan four milestones in one evolving companion: a mobile entry and evidence workflow, a connected journal with AI summary drafts, cited retrieval, and a reviewed mobile agent product. Each milestone needs acceptance evidence before being marked complete.
Can most of the AI work run locally?
Much of the development can run locally. Model suitability depends on available hardware and task quality; an existing hosted assistant is a fallback. Azure exercises may incur charges and need a budget. A deferred deployment stays labelled pending, even when its local equivalent works.
Where can I follow each day’s work?
Every schedule entry opens a dated day page. We plan to publish each chapter one day before its scheduled session. Session results record actual work, tests, and corrections. Reading or publishing a chapter does not mark the exercise or milestone complete.