Available for senior and staff-level roles
Seattle, WASystems Engineer | LLVM/Clang, AI Accelerators & AI Infrastructure

Omar Habra

Systems engineer for high-stakes software.

I work where correctness, performance, and product reality all matter at once: compilers, AI platforms, and hardware-aware systems that have to ship without drama.

Compiler/toolchain internalsAI platform reliabilityPerformance engineeringHardware-aware ML bring-up

Shipped In

Apple, SiFive, Stealth

Platform teams, compilers, and AI product delivery.

Core Domains

LLVM, RISC-V, TTNN

Low-level optimization with evidence-backed regressions.

Infra Range

AWS, Azure, GCP

Multi-cloud AI systems with guardrails and automation.

Work Style

Calm under constraints

Readable systems, fast diagnosis, and strong follow-through.

Selected Proof

Four contexts. Clear evidence.

A concise read of the systems work most relevant to a senior hiring loop.

Constraint

ABI mismatches and codegen regressions were expensive to get wrong.

What I owned

Bridged AST and ABI behavior, validated cross-arch code generation, and built regression tooling.

Impact

Shipped interop work with stronger rollout confidence across x86_64 and arm64.

Tags

SwiftClangABI
Open the Swift PR
// PERFORMANCE_METRICS

Athletics as a systems discipline.

This section matters because it shows how I operate: train deliberately, read telemetry, and keep composure when the conditions get messy.

Why it belongs here

Hiring teams don't need hobbies. They need more evidence about judgment, repetition, and discipline.

Race Telemetry

Real GPS data from actual races

LIVE
Loading GPS data...

Review Habit

After-action review through route traces, splits, and visual footage.

Decision Style

Pace patiently, then push when the data says the system can hold.

Transfer To Work

Preparation, stress tolerance, and clean retros are part of the same muscle.

Performance Dashboard

Marathon build

4 months

Structured block from zero to finish line.

Race format

Olympic

Swim, bike, and run under fatigue.

Combat sport

Purple belt

Adaptation under pressure and constraint.

Training lens

Telemetry

Review the route, not just the result.

Engineering carryover

Measured preparation beats adrenaline. Training blocks look a lot like release plans.
Feedback loops matter. Route files, splits, and footage turn emotion into evidence.
Composure is a skill. Long races and difficult debugging sessions reward the same temperament.
Marathon finisher with friend holding medal

Full marathon

Seattle Marathon

First marathon finish after four months of structured training.

BJJ competitors on podium with medals

Purple belt competitor

Brazilian Jiu-Jitsu

Multiple matches in a single day, adapting under stress.

Athlete at triathlon finish line

Olympic distance

Lake Tye Triathlon

Open water swim, bike, and run.

Public Receipts

The work is inspectable.

Hiring conversations go faster when there are real artifacts to open.

Open Product

DiffSwarm

A multi-agent review CLI that makes one model propose bugs, another challenge them, and a final report synthesize only evidence-backed findings.

Visit product

Compiler Work

Swift-C++ interop

Public compiler contribution shipping getter and setter interoperability inside the official Swift toolchain.

Read PR #40842

Accelerator Work

TT-Metal vision models

MaskFormer, DPT-Large, and YOLOS-small bring-ups for novel hardware, with performance artifacts and implementation detail in the open.

See TT-Metal PRs

Let's Build

If the role needs depth and velocity, let's talk.

I'm especially strong in compiler work, AI infrastructure, product-facing systems, and the messy connective tissue between them.