Software Engineering Intern
Title: Fall 2025 Software Engineering Intern
Requisition: 2025-ENG0010
Category: Hourly, 40 hours/week
Location: Greater Sacramento area, California (On-site/Remote)
Compensation range: $30/hr - $36/hr
About Tiami Networks
At Tiami Networks, our mission is to build a distributed radar system capable of target detection at every point on and above the Earth’s surface using 5G networks. Customer work spans government and Fortune 500 enterprises across three continents. We build integrated sensing-and-comms (ISAC) that ships on hardware-passive/bistatic radar using 5G/LTE/NTN signals of opportunity.
What you’ll do
- Rewrite R&D to production implementations. Turn reference pipelines into maintainable services; optimize planning/reuse, vectorization, memory layout with hardware-aware choices.
- Contribute to GPU inference. Add or adapt CUDA/TensorRT/Triton paths for small models and I/O; start basic and iterate with measured latency/throughput gains.
- Ship and maintain services. Land changes behind flags, add health and metrics, participate in deploys to device-side Docker services, and execute safe rollback when needed.
- Mind the memory. Profile and fix issues across amd64 and aarch64 (heap/stack usage, allocator behavior, copies, queue growth).
Outcomes that define success
- 2–3 weeks: Productionize one R&D module with tests and a micro-benchmark in CI.
- ~6 weeks: Add or improve a TensorRT/Triton path and demonstrate a clear before/after latency or throughput delta.
- ~10–12 weeks: Ship a change that survives an OTA deploy on a lab/field node.
Minimum qualifications
- Proficiency in C/C++ and Python with unit testing and code review experience.
- Working knowledge of FFTs and numerics (convolution modes, windowing, precision trade-offs).
- Basic CUDA or TensorRT or Triton exposure (coursework or project is fine).
- Familiarity with Git and Docker; able to read logs and add simple metrics.
- Understanding of processes/threads, stacks/heaps, and file I/O on Linux.
Nice-to-have (not gatekeepers)
- Experience with SDR toolchains (GNU Radio/UHD/Soapy) or OpenRAN stacks.
- Exposure to Yocto-based embedded Linux, CI, and basic observability.
- Comfort skimming relevant 3GPP sections and tying them to tests.
Tools you might use
C/C++, Python, CMake; FFTW/cuFFT; CUDA, TensorRT, Triton Inference Server; Docker; Git; pytest/googletest; perf/valgrind/heaptrack; simple Prometheus exporters; Wireshark/tcpdump. (SDR targets run Linux/Yocto; services are containerized.)
Practical
- Duration: 10–16 weeks.
- Location: On-site/Remote lab blocks are expected for bring-up and testing.
- Pace: Priorities can shift; scope will stay realistic for an intern.
If you can take a messy R&D block, prove what’s slow, make it faster and safer, and ship it with a rollback plan, you’ll do well here.
Tiami is an equal opportunity employer and prohibits harassment or discrimination of any kind. All applicants will be considered for employment without attention to race, color, religion, sex, sexual orientation, gender identity, family or parental status, national origin, veteran or disability status. This position does not support visa sponsorship, candidates must possess unrestricted US work authorization (CPT/OPT applicants are welcome).
Note to recruitment firms: Tiami will not accept unsolicited resumes. Any such submissions will become the property of Tiami.
