01
AI Feature Delivery
Production LLM features: RAG, agents, copilots, structured extraction, built for real traffic.
Software engineering studio
Ship real software in the age of AI.
Twill Software is led by Rohan Pavone, who spent 7+ years at Google and has a founding engineer's track record taking products from 0 to 1. We help founders and teams design, build, and ship AI-powered software that makes it to production.
Led by ex-Google engineerFree initial consultationOnline or onsite3 client spots this quarter
Services
How we help
Hands-on engineering from the first line of code to production scale and team enablement.
02
0→1 Product Engineering
Founding-engineer capability for greenfield builds across the full stack.
03
Performance, Scale & Cost
Cut API p95 from seconds to ms, separated 1M+ QPS systems, trimmed cloud spend.
04
AI-Assisted Dev & Advisory
Architecture reviews, security and compliance, and team adoption of AI tooling like Claude Code.
05
Team Training & AI Enablement
Onsite or remote workshops that teach engineers to use AI tools and modernize the dev process.
Approach
Senior engineering, start to finish
You work directly with the engineers doing the work. Clear scope and production-grade delivery, with no handoffs and no junior bench.
Book a free intro callBackground
Who you're working with
Twill Software is led by Rohan Pavone. Across 7+ years at Google and as a founding engineer, he has scaled systems to 700M+ records and over a million QPS, cut API latency from seconds to milliseconds, and taken greenfield products from a blank repo to launch.
The practice covers LLM applications, high-throughput APIs, performance and cost optimization, security and compliance, and embedded systems, with fluent use of AI-assisted development.
Rohan: ECE, University of Toronto.
Selected work
- 700M+ multi-cloud security findings ETL (at Google)
- 1M+ QPS re-architecture
- >50% media playback performance gain (Chrome/Fuchsia)
- Autonomous Claude Code agent fleet
- API p95 2.9s → ~300ms
Stack
Tools we work with
Languages
TypeScript Python Go C++ Java SQL
AI / LLM
Claude API OpenAI RAG AI agents Claude Code ML data augmentation
Web & App
Next.js React NestJS Node.js Prisma React-Admin
Data & Infra
PostgreSQL Materialize dbt Terraform AWS Vercel
FAQ
Common questions
How we price, scope, and run engagements. Don't see your question? Ask on a call.
How do you price your work?
Pricing is flexible and matched to the engagement. Instruction and consultation are typically billed as sessions, for example a half-day teaching session. Project work is billed hourly across regular weekly or monthly working sessions. Ongoing engagements run on a retainer, invoiced ahead of the work.
What does a typical engagement look like?
Anywhere from a single half-day session (instruction or a focused consult) to multi-week or multi-month project work with regular check-ins. Ongoing work is structured as a retainer so we stay engaged with your team.
What are your payment terms?
We invoice ahead of the work. Project and ongoing engagements start with a deposit or retainer, with the balance on completion. Individual sessions are invoiced per engagement.
What happens after the free intro call?
We put together a project plan and timeline built around your goals and your team's availability. The plan is yours to keep, a clear and actionable deliverable whether or not we continue together.
How soon can you start?
We usually need about a week to scope and plan a new engagement. To keep quality high, we take a limited number of clients each quarter.
How will we communicate during a project?
However works best for your team: Slack, Teams, or email, paired with a regular check-in call (typically weekly) and a shared doc or board for status and decisions. We keep updates proactive so you always know where things stand.
Who do you work best with?
Founders and teams who need senior engineering they don't have in-house, whether you have limited technical staff, or capable people who lack deep production experience and want a professional finish. Much of the work is data-heavy and about getting the nuts and bolts built right. We can also lead the harder problem-solving when there's no technical team to lean on.
Who will I be working with?
Rohan Pavone leads every engagement and does the core engineering himself. When a project calls for specialized expertise outside his focus, we bring in the right person from a trusted network.
Do you work remotely or onsite?
Remote is our default and the fastest way to get started. If you're in the NJ, NYC, or Toronto area and prefer onsite, we can travel for in-person work.
Who owns the code and IP?
You do. All code and deliverables produced for your engagement are owned entirely by your company.
Will you sign an NDA?
Yes. We're happy to sign your NDA before we get into specifics.
How does team training work?
It scales with your team. We offer a half-day crash course to get people productive quickly, and multi-day sessions for larger groups or deeper enablement, onsite or over Zoom.
Contact
Let's scope your project
The first meeting is always free, online or onsite. Book a time below, or send a message and we'll reply within a day.
Launch offer: the first 5 clients receive a free 2-hour virtual consultation. Mention it when you book.
Schedule a call
Or send a message
Prefer email? rohan@pavoneadvisory.com