Charles Thayer Email & Phone Number

San Francisco, Ca, Us

AI engineering and Infrastructure: Startup, consulting, advising, CTO work, founder squiggle, etc.Talk to me about:* Your infra and operational challenges with AI apps and systems building.* Envoy, Istio, Kubernetes as they related to AI/ML.* AI provenance, citations, RAG for content-driven web sites, web-crawling / web-graph / web-search.* AI production performance related to vector DBs, embeddings, caching, memory and context.

San Mateo, California, Us

Formed the Service Communications team, as tech-lead, which owns networking from app-to-infra. The team runs the api-gateway (envoy) and internal proxies (traefik) for the BEDEV2 clusters (microservices for backend development), owns the network software stack (C#, golang) including gRPC and Protobufs, and provides envoy and service mesh. Also involved with the design, planning, and the move to a cell-based architecture.Worked on the MSP team (Microservice Platform Team) to plan and build the next generation of infrastructure for platform services. Worked on many systems and prototypes related to innovating, scaling, and migrating the architecture (Windows to Linux, monoliths to microservices, bespoke services to open-source, etc.)

Built the tech stack and engineering.Connecting Businesses with customers, using messaging. We've built a managed messaging platform specifically for professionals, out of our experience in Conversational Commerce for The Gap, The Wall Street Journal, many startups, and professionals. SMS, Facebook Messenger, Twitter. Tech Stack: Python/Django/DRF, Redis, Postgres, Kubernetes, Firebase, Twilio. https://stackshare.io/notify-io/notify-io

San Francisco, Us

Built MVP for The Urban Adventure App: http://www.cityscout.us -- Stealth, members-only, not launched. Tech: python, django, bootstrap, nginx, AWS, fabric, puppet, etc.

Large-scale Infrastructure Engineer.Data Infrastructure: Worked on ZippyDB, a distributed and replicated key-value store based on Paxos and RocksDB. Helped port the real-time counter service (unique counters based on HyperLogLog) to ZippyDB (with live customers) and migrated it to Folly Futures for our C++ APIs. Implemented the first production version of the shard load balancer, which enabled balancing on several metrics across possibly heterogeneous hardware.Core Infrastructure: Monitoring Systems: Moved the backend operating metrics and monitoring systems from a MySQL backend to HBase, and worked on Disaster Recovery. Worked on many of the subsystems such as real-time percentile aggregation, prediction modeling, and the real-time computation engine. System processed over 1 Billion time series datapoints per minute at the time I left the team.Heavily involved in debugging many Facebook technical problems (outages / SEVs) and reported at several Incident Reviews, mentored interns, and had a lot of fun at every possible hackathon.

Sunnyvale, Ca, Us

Linux cluster kung-fu.Worked on web search content systems for 7 years, with 4 managers, on 3 teams, during 5 cluster doublings, with thousands of nodes per cluster. Mainly worked on cluster management and monitoring software (in C++) and the operation of the production web crawlers.Content Operations Team: Managed the deployment, operations, and monitoring of production web crawler clusters in distributed data centers. Typical projects included moving 1.8 PB across coasts. Catalog Agility Project: Primary architect on crawler scalability and data storage changes (C++).Crawler Team: Catalog Master: Built and extended cluster management and monitoring software for managing thousands of nodes (C++). Many projects ranging from investigating problems based on statistical anomolies in freshness, comprehensiveness, and quality (Python) to writing disk burnin software (C) used by all of YST (Yahoo Search Technology). Wrote CTA package (Cluster Trends Analyzer) in Python and CluStats (Cluster Stats and alerts) libraries in C++.Echelon Team: Echelon Master: worked on cluster management software for controlling indexing pipeline and batch processing (C++). Echelon Copier (EchCp in C & C++): cross-coast database distribution software with extensive performance throttling controls for delivery production index data to other groups.

Tapstone's Server Change Center is an IT tool for managing enterprises with hundreds to thousands of machines. Agent software running on each computer detects changes to the file system, the registry, and configuration files (on both Solaris and Windows.) These agents each communicate a stream of events in real-time to a central appliance (running Linux and Postgres.) * Designed the distributed network agent for portability, scalability, and low-impact. * Wrote low-level back-end software for Solaris, Linux, and Windows NT in Python and C++. Implemented hardware scanner, control daemon, and various libraries. * Reviewed code, unit tests, system tests, and design work as the manager and lead engineer of the agent group (4-6 developers, including myself.) * Designed APIs, libraries, and protocols as lead engineer of the architecture team during system design phases. Reviewed system-wide design and implementation choices to support the goals of high performance, fail-safe operation. * One of the authors of the provisional patent application: Management and Monitoring System for Distributed Computer Networks. Technologies: Systems Programming (cross-platform); Python, C++, Java; TCP/IP, XMLRPC, SQL.

Built the technology for "The New York City Real Estate Exchange." CityRealty.com was a small on-going venture with new funding, from Goldman Sachs, to create sophisticated web personalization services for residential real-estate brokers, buyers, renters, and customer service representatives. Shared the role of CTO with partner from Mediabridge Infosystems Inc.