Matt Kilmer
AI Product Manager & Software Engineer · Musician/Producer
About
AI product manager and software engineer specializing in LLM-powered applications. I build production systems that solve real problems - from AI-powered content generation at scale to autonomous development workflows. Currently Senior PM at Viewcy, where I lead product strategy for an event ticketing and live streaming platform.
I ship code and understand the technical constraints of working with LLMs: managing context windows, optimizing cost per request, handling non-deterministic outputs, and building evaluation frameworks for subjective quality. I've launched commercial AI products that handle prompt engineering, structured outputs, and function calling - from idea to production.
My approach combines product intuition with engineering pragmatism. I've built AI-powered Shopify apps processing thousands of products, Claude-powered automation tools that create pull requests autonomously, and trading platforms that translate natural language into executable strategies. I know what works in production versus what works in demos.
Background
Before product and engineering, I was a professional musician and producer for over a decade. As Music Coordinator for the Emmy Award-winning TV show "Louie," I shaped the sonic landscape of contemporary television, producing scores with real musicians rather than stock soundtracks. I performed with Grammy-winning artist Lauryn Hill, comedian Reggie Watts, composer Jon Brion, and master oudist Simon Shaheen. I've been a Carnegie Hall Weill Music Institute fellow and produced albums across genres from comedy to classical. That background taught me about iteration under uncertainty, creating experiences that resonate emotionally, and the importance of consistent quality - all directly applicable to building AI products where outputs are probabilistic and user trust is earned through reliability.
Experience
ResumeSenior Product Manager
- •Oversee core product roadmap for event ticketing and live streaming platform
- •Partner with executive stakeholders to define roadmap and set goals in-line with company vision and mission
- •Balance the needs of multiple user classes (artists, fans, sponsors) to drive success for all customers and stakeholders
- •Own product specification and development process, interfacing with engineering, design, UX, marketing, and analytics teams to consistently deliver product releases on time
Founder and CEO
- •Oversee all aspects of business operations and strategy for a boutique eco-conscious health and beauty brand
- •Establish and maintain relationships with vendors and strategic partnerships
Senior Product Manager
- •Led product development lifecycle for flagship music app and interactive digital music service
- •Measured and evaluated product optimizations for business impact including conversion, retention and LTV
- •Hired 10+ employees across product, engineering, sales, and design
- •Business partners included Scooter Braun, Universal Music Group, Warner Music Group
Product Manager
- •Set product vision and strategy inline with business objectives for music software startup
- •Defined releases, oversaw user testing, prioritized features, owned product roadmap
- •Led development of three apps with 1M+ downloads
Music & Production
Music Coordinator
- •Oversaw all aspects of music production, interfaced with the creative team
- •Hired 50+ musicians, managed and delivered assets for all 5 seasons of an Emmy-Award winning television show
Touring / Session Musician
- •Credits include: Lauryn Hill, Reggie Watts, Jon Brion, Ryan Scott, Simon Shaheen, Krishna Das
Projects
Claude Slackbot
Autonomous AI development assistant operating directly in Slack
Production bot that analyzes codebases, generates fixes, creates pull requests, and deploys previews - all from natural language Slack messages.
Production SaaS helping Shopify merchants optimize product listings at scale through intelligent content generation and bulk processing.
TradingGPT
Natural language to executable trading strategies with backtesting
Platform translating plain English into validated, backtested algorithmic trading strategies with risk management and paper trading deployment.
Platform connecting constituents with elected officials through optimized communication workflows and engagement tracking.
Clean Beauty Advisor
AI-powered beauty product ingredient analysis with personalized recommendations
Production SaaS platform analyzing 100k+ beauty products with GPT-4o, helping users make informed choices based on their skin profile and ingredient preferences.
Experimental web application creating real-time particle animations and generative soundscapes responding to user interaction at 60fps.
Building with LLMs
Production AI is Different
Demos optimize for wow moments. Production systems optimize for reliability, cost efficiency, and graceful degradation. The hardest problems aren't the AI - they're context management, quality evaluation at scale, cost-per-request economics, and handling non-deterministic outputs in deterministic systems. I build for production.
Quality Control > Model Selection
The biggest blocker to AI product adoption isn't model capability - it's user trust. I've learned that quality control mechanisms (preview before apply, rollback capabilities, validation layers) matter more than using the newest model. Users need confidence that AI won't break their production systems. Build the guardrails first.
Cost Optimization is Product Design
LLM costs scale with usage, making unit economics critical from day one. I optimize through prompt caching, context window management, structured outputs to reduce parsing, and strategic model selection (using smaller models for simple tasks). Cost-per-request directly impacts product viability - it's not an optimization problem, it's a design constraint.
Evaluation Frameworks Come First
How do you know if your AI product is improving? Traditional metrics don't apply to subjective outputs. I build evaluation frameworks before scaling: human evaluation pipelines, A/B testing infrastructure, quality scoring systems, and failure case analysis. You can't improve what you can't measure, even when outputs are probabilistic.
Safety and Responsibility
AI products carry unique responsibilities. I design with safety constraints: input validation to prevent prompt injection, output filtering for harmful content, rate limiting for abuse prevention, audit logging for accountability, and clear communication of AI involvement to users. Building responsibly means anticipating misuse, not just optimizing for the happy path.
Skills
AI/ML
- •LLM Integration & API Design
- •Prompt Engineering & System Design
- •Context Window Optimization
- •Structured Outputs & Function Calling
- •Streaming & Real-time Processing
- •Prompt Caching & Cost Optimization
- •Evaluation Frameworks
- •Safety & Content Filtering
Product
- •AI Product Strategy
- •0→1 Product Development
- •Roadmap Planning
- •User Research & Testing
- •Stakeholder Management
- •Go-to-Market Strategy
- •Agile/Scrum
- •Product Analytics
Technical
- •TypeScript/JavaScript
- •React/Next.js
- •Node.js
- •Python
- •PostgreSQL
- •API Design
- •AWS/Vercel
- •Git/GitHub