Curriculum vitae
Vitalik Garan
Senior Technical Program Manager · AI-Native Platforms · ex-Amazon Ring · Alexa · Wolt
Summary
Senior Technical Program Manager who builds with AI, not just manages it. I connect systems that don't talk to each other -- technical, human, or both. 10+ years shipping complex programs across AI/ML, IoT, and hardware. A radar camera to 800k-2M units at Ring. Speech recognition to production at Alexa scale. An AI workflow system built from scratch at Wolt. This year -- Kestrel, a production AI platform shipped and operated solo, with evals in CI (93.6%-recall scoring filter, 277-item human-judged benchmark) and its AI cost cut from $6.51 to $0.84 per run. The pattern -- take the ambiguous, cross-functional, hard-to-own problem and find the path through it.
Experience
Builder & Operator -- Kestrel (AI Job-Market Platform)
- A deliberate choice after Wolt: went deep into building, learning, and living inside the 2026 multi-agent AI stack instead of watching it from a deck -- now running my own work as an orchestrated fleet of AI sub-agents.
- Working alone for half a year, got promoted to Builder, Product Manager, QA, CFO, and CTO of myself.
- Designed, shipped, and operate Kestrel solo (FastAPI + React, ten AI providers, 4,600+ tests, PyPI/AGPL) -- evals run in CI: scoring filter measured at 93.6% recall / 74.6% precision on a 277-item human-judged benchmark; AI cost per run cut $6.51 to $0.84.
- One of screenpipe's most active external contributors -- 24 merged upstream PRs + a dev/test harness I dogfood 24/7 -- work that led to a direct offer from the founder.
- Looking for a team, a cause, and a place to call home: philosophically, existentially, operationally.
Technical Program Manager
- Joined to build the AI-augmented program function from scratch in a post-M&A environment with no established tooling culture.
- Built AI-powered TPM intelligence system -- 6 AI tools (Claude, OpenAI, Gemini) integrated with 4 enterprise APIs (Atlassian, GSuite, Slack, Glean) -- turning hours of manual status collection into single-prompt program synthesis.
- Managed Salesforce CRM transformation across product, engineering, and business stakeholders.
Technical Program Manager -- ML, Speech Recognition & LLMs
- Ran continuous ASR training and deployment pipeline across French and Italian locales -- research teams in Aachen/Bangalore, engineering in Seattle.
- Cut ASR release response SLA from multi-day to 1 business day via bi-weekly deployment redesign and ownership clarity.
- Solved the classic ML org problem -- got researchers who want to experiment and engineers who need to ship into working alignment across 3 continents.
- Drove early LLM integration initiative alongside continuous ASR work during the inflection year for large language models.
Technical Program Manager -- Sensor Research
- Led Ring Ultra radar camera from sensor research to 800k-2M units shipped -- 10+ teams, 5 countries, 4 years, spanning ML, hardware, firmware, and software simultaneously.
- Shipped Motion Settings 2.0 to 40M+ Ring devices in 1M/day controlled waves -- drove Person Detection from 30% to 46% on deployed hardware; managed full platform rollback after accidental global deploy, COE completed, feature re-shipped correctly.
- Ran ANVIL compliance overhaul -- split monolithic Ring Research AWS account into 25 independently certified services against Amazon's highest data and privacy standards, zero downtime, 5 certifications 30 weeks ahead of schedule.
- Made the uncomfortable call -- terminated a high-performing but toxic engineer; team cohesion and delivery predictability measurably improved.
- Delivered portfolio of 7 products fusing ML, hardware, firmware, and software domains.
Fleet & Field Program Manager
- Built 5 modded VW demo cars deployed across USA, Europe, and Japan -- 200+ live demo drives, sales engagement up 5x vs. slide decks.
- Landed several $1M+ contracts directly attributed to the demo car experience.
- Reverse-engineered VW CANBUS with C++ engineers to pull native sensor data into the product stack.
- Cut demo prep from 1 month to 1 week by integrating field team directly with R&D.
QA Engineer, then Product Owner -- Automotive & Cloud
- QA Engineer first -- built and automated test suites (Python, bash, Selenium) across 5 platforms for a Japanese automotive OEM and a cloud storage product, to 100% coverage.
- Then Product Owner for cloud storage and an automotive fleet-management portal -- turned 1,500 PDF specs into a live Axure prototype for remote fleet management, cutting development time 85%.
The career runs further back: production and licensing at Room 8 Studio (gamedev -- Piano City hit #1 on the iOS App Store in 126 countries), and corporate PR at Mykhailov & Partners.
Projects
- Kestrel -- my meta-job-search effort; the interesting half is the scraping and multi-strategy scoring across providers, not the CV generation. Solo: FastAPI + React, ten AI providers behind one interface, an eval harness, 4,600+ tests, PyPI/AGPL. Case study.
- screenpipe -- one of the most active external contributors to the YC-backed local-first capture tool: 24 merged upstream PRs (privacy/PII redaction, CoreAudio + meeting-detection fixes, contributor onboarding + a dev/test harness) and 40+ issues filed. I dogfood my own build 24/7 and contribute the slow way -- a real local harness, not a vibe-coded PR.
- web-agent-comparison -- benchmarked 7 browser-automation MCP servers against frozen fixtures and surfaced real vendor bugs (30s hangs, phantom listeners, TLS-fingerprint leaks).
- awesome-llm-token-optimization -- a curated guide to cutting LLM cost (batching, caching, provider routing); the playbook behind running Kestrel cheap.
- llm-safe-haven -- published npm tool for one-command security hardening of AI coding agents.
- chronoclaude -- maintained derivative shipped as my own idle-timing plugin with an MCP time-tools server (v0.5.4); started from a maintainer-reviewed upstream proposal.
Education & Certifications
MS, International Business Management
- Software Product Management Specialization -- Coursera (2017).
- Cambridge ESOL Certificate in Advanced English (CAE).
Skills
- Program Management
- Cross-functional leadership, SDLC, Agile/Kanban, stakeholder management, risk mitigation, OKRs, roadmap planning
- AI & ML
- LLM integration, MCP architecture, agent systems, prompt engineering, RAG, AI-augmented workflows, Cursor, Claude Code
- Technical Domains
- IoT, computer vision, speech recognition (ASR), radar/sensor R&D, embedded systems, AWS infrastructure
- Tools
- Jira, Confluence, Linear, GitHub, Looker/Tableau, Salesforce, Slack APIs, GSuite APIs
- Languages
- English (fluent, CAE C1), Ukrainian (native), Russian (native), German (learning)