Selected work
Shipped.
Shipped.
Live, in use.
We don't do mockups. Every project below is in production, used by real people, and was delivered start-to-finish in days - not quarters.
Product · 14-day build
SubmitX
Chrome extension + AI agent that tailors your resume per role, fills the application form, and submits. 30,000+ submissions every day across Workday, Greenhouse, Lever, Ashby and 47 ATSes.
- 30,000+ / day · live submissions across the fleet
- 5 days · idea to first live submission
- 10+ microservices · one repo, Lambda-deployed
- $0 committed cloud spend · pay-per-use only
- Stack: Next.js 16, Clerk, Dodo Payments, Anthropic Claude, AWS Lambda
Capability · In-house LLMs
Self-hosted LLMs in your VPC.
Your own model, in your own network. We fine-tune Llama, Qwen and Mistral, then deploy them on vLLM, TensorRT-LLM or SGLang. Latency you control, cost you predict, data that never leaves your perimeter.
- Train: LoRA, QLoRA, DPO, GRPO on Axolotl / Unsloth
- Serve: vLLM, TensorRT-LLM, SGLang with FP8 + speculative decoding
- Deploy: H100, L40S, your on-prem GPUs - air-gapped on request
- Observe: Langfuse, OTEL traces, evals as code, replayable failures
More coming
Your project, next on this page.
We take on a handful of projects each quarter. If you have a problem worth shipping fast, we want to hear about it.
Start a project