luis.samaniego.engineer

Luis Samaniego

Software Engineer·AI-driven engineering workflows·Go, Python, cloud

I build backend systems and the tooling around them, and for the last two years I have been designing how AI agents can do that work reliably — specs, decisions and checks included.

NOWShipping Pitwall — open source soon
GitHubLinkedInEmailResume PDFSOON
BASE
Mexico
STACK
Go · Python · Cloud
FOCUS
Backend · AI workflows
ROLE
Senior Implementation Engineer
LOG
2016 → now
01

Featured

CASE STUDY/ SELF-HOSTED / AI AGENTS

Pitwall

A self-hosted control room that drives headless Claude Code agents through a deterministic pipeline, with every step visible from one web UI.

108ADRS
20MILESTONES
482COMMITS
SAMPLE RUN · 7/7 GREENTOTAL 24:09 · 263.4k TOK
  1. S1
    enrich
    0:48
    6.2k tok
  2. S2
    plan
    2:15
    21.4k tok
  3. S3
    implement
    11:32 · 2 passes
    148.0k tok
  4. S4
    code-review
    3:04
    32.7k tok
  5. S5
    qa-code
    4:21
    41.0k tok
  6. S6
    release
    0:57
    4.8k tok
  7. S7
    qa-notes
    1:12
    9.3k tok
Visit pitwall.samaniego.engineerSOONGitHubSOON
enrich → plan → implement → code-review → qa-code → release → qa-notes
02

How I work

5 rules · applied to people and agents alike
  1. P1

    Specs before code

    Scope and out-of-scope are written before the first commit.
  2. P2

    Decisions recorded as ADRs

    Every choice keeps its why, so nobody re-argues it.
  3. P3

    Small phases, one commit each

    Work lands in reviewable steps, never one giant diff.
  4. P4

    Gates an agent cannot skip

    Review and QA verdicts decide what moves forward.
  5. P5

    Checks a human can verify

    What tests can’t prove becomes a manual check.