Vision Model Acceleration
Tenstorrent2025 (Bounties)
Open-source bounty work bringing up DPT-Large depth estimation on Wormhole N300 accelerators using TTNN, with validation and performance profiling.
Omar Habra • B.S. Computer Engineering • Compilers & AI Systems • Ultramarathon & BJJ
Built like an athlete.Shipping compilers and AI systems that hold up in production.
I work across compiler/toolchain engineering and production AI infrastructure: interoperability, performance tuning, and reliability guardrails. Outside of work I train for ultramarathons and BJJ, which keeps me consistent—but the work stands on its own.
- Shipped Swift/C++ interoperability work at Apple, contributing to the Swift/Clang toolchain.
- Tuned LLVM/Clang for RISC-V RVV vector workloads at SiFive and added regression coverage for performance-sensitive changes.
- Contributed TT-Metal TTNN model bring-ups (MaskFormer, DPT-Large, YOLOS-small) and built production AI systems for clients.
Runtime Profile● Active
Compiler / IR95%
GPU kernels90%
Distributed systems80%
ML infrastructure85%
Endurance Profile▲ Training
Longest race
100mi
Mountain ultra
Elevation
+6,000m
Total gain
Fastest 100k
9h 30m
Weekly vol.
80-100km
Currently focused onLive
Swift/C++ interoperability and LLVM performance
Trail running in the Pacific Northwest
Same endurance mindset as debugging production systems
Scroll to explore
About / Why work with me
Compilers and AI systems, built for production.
Systems engineer with a B.S. in Computer Engineering, spanning compilers/toolchains and production AI systems.
Built and shipped compiler features at Apple (Swift–C++ interoperability, Apple Silicon bring-up) and SiFive (LLVM/Clang optimization for RISC-V RVV).
In consulting, I design and implement secure multi-tenant LLM products on AWS—retrieval pipelines, evaluation/guardrails, and deployment automation.
Recent open-source work includes TT-Metal TTNN model bring-ups for MaskFormer, DPT-Large depth estimation, and YOLOS-small object detection.
Systems instincts
I optimize for correctness, debuggability, and operational reality: tight feedback loops, good tests, and the instrumentation you need when things go sideways.
Compiler depth + shipping
Comfortable in LLVM/Clang internals and equally comfortable getting things into production: infrastructure as code, CI, and careful rollouts.
Clear, pragmatic collaboration
I write down decisions, communicate tradeoffs early, and do the unglamorous glue work that keeps teams moving.
Currently
Building and shipping work across toolchains, production AI infrastructure, and accelerator bring-ups—while keeping a steady training habit outside of engineering.
Featured Work
Technical contributions with measurable impact
From compiler infrastructure to AI acceleration, each project represents production-grade engineering with real-world results.
RISC-V Vectorization
SiFive2022–2023
Significant performance improvements on NASA benchmark suite through LLVM vectorization tuning and custom cost models for RISC-V Vector Extension.
Swift-C++ Interoperability
Apple2020–2022
Led the implementation of C++ getter/setter support in the Swift compiler, enabling seamless property-style access across language boundaries. Featured in WWDC 2023.
Open Source
Contributions to compilers and AI acceleration
Swift compiler work shipped in the public toolchain. TT-Metal contributions bring up modern vision models on Tenstorrent accelerators.
Experience
Building at the cutting edge
Building compilers, toolchains, and production systems at teams pushing the boundaries of what software can do.
Independent Consulting
AI Infrastructure & Systems Engineer
2023 – Present
Seattle, WA
- Built and shipped a production tutoring workflow: document ingestion → embeddings → retrieval (Postgres/pgvector) → Bedrock/Claude orchestration → Next.js UI.
- Designed secure multi-tenant AWS architecture with per-tenant isolation, RBAC, Secrets Manager, VPC networking, and audit logging; delivered infrastructure as code with Terraform.
- Added reliability guardrails for LLM features: structured outputs (Pydantic), prompt/tool validation, regression tests, and monitoring to keep answers stable across curricula.
SiFive
Senior Software Engineer
2022 – 2023
San Mateo, CA
- Improved performance for RISC-V vector workloads by tuning Clang/LLVM vectorization cost models and codegen heuristics for the RISC-V vector extension (RVV); added microbenchmarks and regression coverage to prevent backslides.
- Integrated LLVM analysis tools into FreedomStudio; built Python automation for toolchain builds and custom utilities for RISC-V ISA analysis.
- Built CI for cross-compilation, ISA validation, and performance benchmarking; automated vendor environment setup with Ansible.
Apple
Software Engineer — Clang/Swift Compiler
2020 – 2022
Cupertino, CA
- Worked on Swift–C++ interoperability across the Swift/Clang toolchain: AST bridging, name mangling compatibility, and ABI integration; contributed to Swift Evolution and open-source compiler infrastructure.
- Supported Apple Silicon (arm64) compiler migration: validated code generation, resolved ABI compatibility issues, and improved Clang/LLVM backend behavior for M1-class targets; partnered with internal product teams for rollout.
- Built compiler reliability tooling: CI bots across x86_64/arm64, automated log triage, and regression detection for Clang/Swift/LLVM issues.
Technical Expertise
Deep knowledge across the stack
From low-level compiler internals to production AI infrastructure, with a focus on performance and reliability.
Compilers & Systems Engineering
LLVMClangSwift compilerRISC-VCode generationVectorization+4 more
Languages & Low-Level Skills
C++PythonSwiftTypeScriptJavaScriptAssembly+2 more
AI/ML & LLMs
RAGLangChainpgvectorEmbeddingsAWS BedrockClaude+8 more
Product & Web Development
ReactNext.jsNode.jsGraphQLREST APIsARKit+3 more
Cloud & Infrastructure
AWS ECS FargateAWS Step FunctionsAWS LambdaAWS S3AWS RDSTerraform+4 more
Databases & Storage
PostgreSQLpgvectorRedisS3Vector databases
Security & Compliance
Multi-tenant architectureRBACData isolationAWS Secrets ManagerAudit loggingPrivacy-by-design
Observability & Reliability
MonitoringTracingCost optimizationIncident responseSLA managementLoad testing
Education
San José State University
B.S., Computer Engineering • San José, CA
Languages
Arabic — NativeEnglish — Fluent
Beyond the Keyboard
Endurance builds discipline
Endurance athletics teaches the same lessons as systems engineering: deliberate practice, measurable progress, and staying calm under pressure.
Seattle Marathon 2024
26.2 miles through downtown Seattle
Loading GPS data...
Full marathon
Seattle Marathon
First marathon finish after four months of structured training.
Purple belt competitor
Brazilian Jiu-Jitsu
Multiple matches in a single day, adapting under stress.
Olympic distance
Lake Tye Triathlon
Open water swim, bike, and run.
Get in Touch
Let's build something together
Open to compiler engineering, AI infrastructure, and systems architecture opportunities. Based in Seattle, WA, remote-friendly.
Ready to discuss your next project?
Whether you need compiler expertise, AI infrastructure, or performance engineering, I'd love to hear about what you're building.