AI and infrastructure engineering for SaaS teams
Kempbell Consulting builds and fixes production systems: LLM features, ML pipelines, AWS infrastructure, CI/CD, and cloud costs.
Run by Travis Kempbell, a Washington-based principal engineer with 15 years of experience. Every engagement is delivered by him directly.
Problems I Solve
SaaS companies hit predictable problems as they scale. Here's how I help.
AI feature not ready for production?
I build LLM features with latency budgets, per-request cost tracking, and evaluation before release. I operate a commercial AI product, including its custom speech-recognition engine, and I bring that experience to your stack.
Overpaying for AWS?
Audit for unused resources, right-size instances, and get inference and GPU spend under control. Past engagements cut 30% or more from monthly bills.
Deployments constantly breaking?
CI/CD pipelines with proper testing, staging environments, and rollback mechanisms for smooth, predictable deployments.
Compliance audit headaches?
Infrastructure-as-code practices and security controls that satisfy SOC2 and data-protection audits without slowing down development.
How I Help
Three engagement types. All work is scoped up front and delivered by Travis directly.
Cloud Cost & Performance Audit
2 weeks
A two-week audit of your AWS spend and performance, including inference, GPU, and per-request costs. Past engagements cut 30% or more from monthly bills.
- Complete AWS account audit covering hidden costs and inefficiencies
- AI workload costs: model serving, GPU utilization, per-request economics
- Includes hands-on implementation of the findings
- Automated cost tracking and monitoring left in place
- 30-day follow-up to confirm sustained savings
Fractional Platform Engineering
3 months minimum
A principal engineer on retainer: architecture reviews, infrastructure management, and hands-on work, with response windows agreed up front.
- Proactive infrastructure management and architecture reviews
- CI/CD pipeline maintenance and improvements
- Async support with defined response windows
- Monthly infrastructure reviews and sprint planning
- Developer tooling and automation
AI Feature Build
4-6 weeks
LLM features built into your existing product: chat, structured extraction, semantic search. Includes latency budgets, cost controls, and evaluation before release. The alternative is hiring an ML team: $400-600K a year and months of ramp before anything ships. This is a fixed quote, delivered in weeks, with help hiring the one maintainer it needs afterward.
- LLM integration built for production: latency, cost per request, reliability
- Structured extraction and semantic search over your data
- Evaluation and A/B testing before release
- Built into your existing stack (Next.js, TypeScript, Node.js, Python)
- Documentation and handoff your team can own
Every engagement is scoped individually and quoted in writing. Engagements typically start at $10K, with retainers from $6K/month.
More about the practice: read about Travis.
Experience & Results
Documented results for growing SaaS companies.
$160K+ in documented annual AWS savings.
Saved one client over $100K/year in AWS spend
Right-sizing, cleanup of unused resources, and reserved instance planning cut the monthly bill with no performance loss. Details in the case studies.
Builds and operates a commercial AI product
A live LLM-based SaaS with its own custom speech-recognition engine (NVIDIA Parakeet) and real-time inference pipelines on ONNX and CUDA. Built and operated end to end.
Led infrastructure through SOC2 and DPA audits
Established secure infrastructure practices and documentation that helped companies pass rigorous security audits without disrupting development workflows.
15 years across AWS, Linux, and CI/CD systems
Principal-level expertise across the full stack: infrastructure-as-code, containerization, monitoring, automated deployment pipelines, and the application code that runs on top.
Technical Expertise
AI & ML Systems
- LLM integration
- Speech recognition (ASR)
- Inference optimization
- Model evaluation & A/B gating
Infrastructure & DevOps
- AWS
- CI/CD
- Linux
- Cloud cost optimization
Full-Stack Development
- React & Next.js
- Node.js
- TypeScript
- API design
System Design
- Latency optimization
- CDNs
- Distributed workloads
- Scalable architectures
Every engagement is delivered by Travis directly. When work calls for skills outside his scope, he brings in trusted specialists and stays accountable for the result.
Get in Touch
If you have an AI feature that needs to reach production, an AWS bill that needs attention, or infrastructure that needs an owner, get in touch.
The first step is a 30-minute call to go over your situation and see if it's a fit.
Email replies within one business day.
What to Expect
- A short call to understand your constraints and goals
- A written proposal with scope, timeline, and a quote
- All work delivered directly by Travis