Looking for a full time Go position

LOCATION: Dakar, Senegal (UTC+0 — full overlap with Europe, morning overlap with US East)
REMOTE: Yes. Remote only, worldwide. No visa or relocation needed; I work as a
contractor or through an EOR, whichever is easier for you.
WILLING TO RELOCATE: No
TECHNOLOGIES: Go (9y), JavaScript/TypeScript (10y), MongoDB, PostgreSQL,
WebSockets, Docker, LLM agents, MCP, static analysis (go/ssa)
RÉSUMÉ: https://www.linkedin.com/in/cheikh-seck-2b449b26b/
EMAIL: cheeikhseck@gmail.com

I’ve been writing Go since 2017 and shipping backend systems since 2014. Most of
it has been the unglamorous kind: warehouse management backends, parcel tracking,
banking integrations, real-time dashboards over WebSockets, PDF and spreadsheet
generation, Twilio/WhatsApp/Outlook integrations. Systems that had to stay up and
be correct, not systems that had to demo well.

For the last two years I’ve been building AI tooling in Go, because that’s where
the interesting backend problems moved:

  • godex — an AI coding agent in Go.
    Terminal-first, tool-calling loop, context management.
  • lite-agent — a browser side-panel assistant that operates on the page you’re
    already looking at, instead of a separate chat window.
  • leakcheck — a taint analyser for Go built with Go . Walks
    SSA, classifies guards into eight classes, tracks taint through interprocedural
    summaries. The honest result was that only 3 of 14 guard-shaped functions
    actually cut taint.
  • An EBNF-based validator for LLM tool calls, built with Go.
    Rejects malformed function calls before they execute instead of retrying them.
  • A Go-native inference engine — matrix multiplication, SIMD paths on Go 1.27,
    no CGO.

What I’m good at: reading unfamiliar codebases, making a service observable and
boring, and finishing things. I was a founder from 2014 to 2017, so I’m
comfortable with ambiguity and with owning a problem end to end rather than a
ticket.

What I’m looking for: a backend or platform team where Go is the default and
where AI work is real infrastructure rather than a demo. I’d rather be the person
who makes the inference path faster or the agent loop more reliable than the
person who writes prompts.

I write about Go regularly at link:cheikhhseck.medium.com/ — mostly measured
teardowns of things I built, including the parts that broke.

Happy to do a paid trial task. Email me.