Let's Talk
San Francisco, CA, USA
Open to the right opportunities
What actually moved

Things that mattered.

Scaled an API platform to 2B+ daily requests and 40k peak RPS, then led a zero-outage gateway migration with ~80% annual cost reduction.
Indeed, Inc.
Built a human-in-the-loop data collection platform for hundreds of global workers, with task routing, dynamic forms, performance tracking, and quality checks.
Indeed, Inc.
Built durable AI workflow infrastructure: customer integrations, Temporal workers, Git mirror throughput, observability, and multi-tenant execution foundations.
Pi Security
Built startup infrastructure from scratch across auth, billing, Redis/FastAPI services, Terraform, GCP, and production observability.
Ventrilo.ai
Chapters

The last decade.

2025 — PRESENT
Founding Engineer, AI Workflow Infrastructure
Pi Security
Agentic cybersecurity workflows • San Francisco, CA, USA
• Built customer integrations across GitHub, GitLab, Bitbucket, Jira, Linear, Notion, Confluence, and related systems.
• Built horizontally scalable Temporal workers for durable agent jobs, data ingestion, webhook-triggered workflows, and long-running execution.
• Built high-throughput Git mirror infrastructure using bare repos on ephemeral NVMe disks, daily tarball backups, and custom tarball advertisements.
• Implemented LGTM observability with Langfuse trace routing for debugging and operating agentic workflows.
Current focus
2023 — 2025
Founding Engineer, Product Infrastructure
Ventrilo.ai
Menlo Park, USA • B2B → B2C
• Built production-critical auth, billing, Redis/FastAPI services, GCP infrastructure, Terraform, and observability from scratch.
• Shipped context-aware AI autocomplete at p95 ~500ms for thousands of users worldwide (browser extension).
• Built realtime context ingestion, RAG post-processing, CRDT/Yjs chat sync, Stripe billing, and realtime JWT revocation via PubSub.
• Prototyped browser activity capture, replay, and distributed crawling systems for product exploration and test intelligence.
Public launch April 2025
2019 — 2023
Staff Software Engineer, API Platform
Indeed, Inc.
San Francisco, CA, USA
• Helped grow GraphQL from one internal client into a standard platform spanning 250+ services and 1000+ API clients.
• Led full migration to a new API Gateway (vendor evaluation → contract → cutover) on a platform serving 2B+ daily requests at 40k peak RPS. Result: ~80% cost reduction, zero outages (previously every 3–6 months), dramatically richer telemetry.
• Co-owned migration to Apollo Federation v2 supporting 250+ services and 1000+ clients.
• Built internal AI knowledge base (OpenAI embeddings + LangChain/LlamaIndex) used by the engineering organization.
• Mentored engineers, ran 300+ technical interviews, led design reviews and major outage investigations.
2015 — 2019
Software Engineer, Data Collection and Crawling
Indeed, Inc.
Tokyo • Web crawler ecosystem
• Built a crowdsourcing platform for collecting and validating long-tail job posting leads across regions and languages.
• Implemented task subscriptions, ranked next-task delivery, dynamic forms, assignment and task routing, worker identity, performance tracking, and downstream ingestion triggers.
• Added quality-control mechanisms including majority voting, cross-validation, and consistency checks before data entered crawling pipelines.
• Enabled hundreds of global workers to complete hundreds of thousands of tasks, contributing to >$10M in customer LTV.
• Built a websites classifier on the publicwww.com dataset — ~750k positively identified sites fed into crawler discovery.
2009 — 2015
Earlier chapters
Yandex
Web analytics (Metrica) • Java / ClickHouse
Mail.Ru Games
MMORPG mechanics (Allods Online) + high-volume crawling platform that delivered >$10M customer LTV
Mail.Ru Group
kino.mail.ru / afisha.mail.ru
GE Money / Mirantis
Helpdesk systems, batch processing, large-scale data work (during university)
How I think about the work

Craft over flash.

Workflow reliability
Durable execution for long-running jobs
Observable pipelines with clear failure modes
Systems that survive nights, weekends, and growth
Operational foundations that reduce firefighting
Data quality systems
Human-in-the-loop task routing
Cross-validation and consistency checks
Dynamic forms shaped by prior choices
Downstream ingestion that can be trusted
Platform maturity
Terraform, IAM, auth, billing, observability
On-call-ready systems and incident discipline
Migration paths that reduce risk
Telemetry that makes production explain itself
Startup execution
Prototypes that clarify product direction
Big deliverables owned end to end
Mentoring without slowing the team down
Comfort navigating vague, early-stage work

Right now.

Building reliable infrastructure for AI workflow systems: durable execution, integrations, codebase access, observability, and operational foundations.

The work I care about turns fast-moving prototypes into systems teams can trust without living in constant firefighting mode.
The foundation

Education &
continuous learning.

Specialist in Mathematics (equivalent to MS)
Computer Security & Cryptography • Saratov State University • 2012
Deep learning & ML
Deep Learning Specialization — Andrew Ng
Machine Learning with Python — IBM
Scalable ML on Big Data (Spark) — IBM
CS231n (Stanford) — Karpathy et al.
Algorithms & languages
Algorithms I & II — Princeton (Sedgewick)
Functional Programming in Scala — Odersky
Kotlin for Java Developers — JetBrains
Blockchain & decentralized systems
Bitcoin & Cryptocurrency Technologies — Princeton
Blockchain Specialization — University at Buffalo
DeFi Specialization — Duke
Blockchain and Money (MIT 15.S12) — Gary Gensler

Let's build something ambitious.

alexlopashev@gmail.com
San Francisco, CA, USA • Open to the right opportunities