Lead, inbox, task, and reporting automation
Connect sources like Gmail, Sheets, Notion, Slack, Telegram, CRMs, and HTTP APIs into one controlled workflow.
Profile
Research background, founder mindset, automation delivery.
My work sits at the intersection of quantitative research, machine learning, operations, and practical automation. I am useful when a team needs someone who can reason about data, build a workflow, and keep the system understandable enough to operate.
Head of Research, Wunder Fund
Machine learning, neural networks, research systems, and HFT talent conversations.
Founder, Accurato
Operations management, analytical systems, and founder-level execution.
Co-Founder, Summator.me
Product thinking around summaries, research, and structured information flows.
Co-Founder, Onederx
Building and operating products where data, process, and user workflows matter.
What I build
Useful automation for teams that are tired of copy-paste work.
The goal is not to sell you a mysterious autonomous agent. The goal is to find one process where automation can reliably save time, then design it so your team can inspect, approve, and improve it.
Data extraction, validation, and reporting
Use SQL thinking to structure messy operational data, validate outputs, and create reports that can be trusted.
Workflow architecture your team can inspect
Visual nodes, credentials, retries, logs, and human approval steps instead of opaque scripts nobody wants to own.
How I think
Automation should be measurable, reversible, and boring in production.
I focus on the parts that usually decide whether an automation survives after the demo: data quality, failure handling, logging, approval flow, and the line between useful assistance and unsafe autonomy.
TriggerEmail, form, webhook, schedule, CRM update.
ExtractPull the fields that matter and normalize the data.
ReasonScore, classify, summarize, route, or draft.
ValidateCheck missing fields, duplicates, and risky outputs.
ApproveHuman review before send, pay, delete, or confirm.
LogExecutions, errors, outcomes, and next actions.
Best first use cases
Start where time loss is obvious.
Lead research
Enrich leads, score fit, write outreach drafts, and update the CRM without manually jumping between tabs.
Inbox triage
Classify incoming messages, extract tasks, route urgent items, and create structured follow-ups.
Meeting to action
Turn transcripts into summaries, decisions, owner-based tasks, and follow-up drafts.
Support drafts
Classify tickets, retrieve context, suggest replies, and escalate cases that need human judgment.
Research monitoring
Track sources, summarize changes, organize findings, and send a digest for finance, crypto, recruiting, HFT, or founder research.
Automation sprints
Two clear ways to start.
Use the blueprint if you need clarity before implementation. Use the pilot if you want the first workflow live.
SQL Automation Blueprint
A focused one-week engagement to identify the best automation opportunity and design the first workflow.
- 60-90 min workflow discovery call
- Automation opportunity map
- ROI, simplicity, and risk prioritization
- Detailed blueprint for one high-value workflow
- n8n architecture plan and risk review
- 30 min follow-up call
SQL Agent Pilot Build
Everything in the blueprint plus one working n8n-based automation connected to your tools.
- One working workflow or agent-style pilot
- Up to 2-3 integrations
- Prompt, tool, and approval-flow design
- Basic error handling and logs
- 60 min handover session
- 7 days async support for small fixes
Who should write
If your team is doing the same operational task every week, there is probably a first workflow worth testing.
- You already use tools like Sheets, Gmail, Slack, Telegram, Notion, Airtable, HubSpot, or a custom API.
- You have repetitive messages, reports, summaries, research, lead handling, or ticket triage.
- You want automation, but you still need approvals, logs, and a clear fallback when something fails.
- You prefer a scoped pilot over an open-ended consulting project.
Send me one messy workflow. I will tell you if it is worth automating.
The fastest way to start is simple: describe the manual process, the tools involved, and what a good outcome would look like. I will suggest whether a blueprint or pilot build makes sense. Telegram: @mr_vvv. If you are an HFT quant, message me directly as well.