Chinh (lelouvincx) / Weekly Report - 2026-W12

Created Mon, 25 May 2026 00:00:00 +0000 Modified Mon, 25 May 2026 06:02:25 +0000
1195 Words
page

Weekly Report — Mar 16 – Mar 22, 2026 (friday mode)

Work

Internal

  • DONE Migrate Census to Fivetran — Completed migration of reverse-ETL syncs (BigQuery → HubSpot) from Census to Fivetran Activations before the April 1 deadline. Two syncs migrated: dim_contacts → HubSpot Contact and fct_customer_stats → HubSpot Company. Validated MAR stays within free tier limits (~1,215 records/month vs 3,500 limit). Follow-up noted for next week: check Census usage estimate to confirm free tier eligibility.
  • DONE Review monitoring dashboard DE-187 — Reviewed PR #60: “Add subscription monitoring dataset and dashboard for customer identity tracking” (merged).
  • DONE Check #data-ops-bot on viz_settings from PR #842 — Fixed duplicate snapshot names for viz_settings (PR merged Mar 17).
  • DONE Continue commenting the project DE-201 (PR Workflow enhancement) — Finished laying comments on the PR Workflow spec. Focus was on things needed during PR preview to facilitate reviewing; anything that doesn’t help confidence in approving a PR is out of scope.
  • DONE 1-on-1 meeting with manager (anh Triet) — Key concerns discussed: (1) integrating AI into the data team’s development experience, (2) current data infra/pipeline complexity and trust issues, (3) PR Workflow adoption. Wrote up 1-on-1 report.
  • DONE Understand the MRR data flow — Clarified that the revenue planning logic is produced from DTS100 and DTS106. Confirmed it is adequate to reproduce these sheets inside Holistics, but will temporarily defer in favor of adding a sales motion dimension first. Next step: ingest Calendly data into the warehouse.
  • IN-PROGRESS Check failed dbt tests & create Linear issues — DAT-554 fixed (Elementary dbt test n_records column missing, PR #841). DAT-555 still open: fct_job_queue_performance model is stale and likely to be removed; mart_product__dataset_datamodel_dimensions also under review. Several skipped downstream models identified: itg_customer_support__organizations, mart_customer_support__organizations, mart_product__activities, and others.
  • IN-PROGRESS Think about applying AI into resolving #data-ops-bot issues — Drafted an approach: pick a task → resolve manually → write detailed guide → let agent attempt → iterate. Then materialize as a reusable skill. Delegation strategy: routine dbt test fixes can go to juniors (Thuan) or agents.
  • TODO Prepare for phase 2 MRR problem — Upcoming. First step: ingest Calendly data.
  • TODO Ingest Calendly data — New task, supports the MRR analysis.
  • TODO Think about how a normal data team would adopt a BI chatbot — Exploring candidates (e.g., nao).

Presales

  • DONE Onboarding call with Showbie — Completed Mar 13.
  • DONE Debrief the last call with Showbie — Key learnings: try out Calculation Builder for immediate use in calls; identified need to improve call quality by studying past calls led by Huy/Vincent.
  • DONE Showbie use case — embedded dashboard QA — Scoped requirements: Showbie PM needs a staging/approval workflow for embedded dashboards. Discussed with team (anh Tai / Huy) and decided on the approach: PM previews via Development → Embed Portal link; dashboard changes on a branch reflect in the embed preview without needing a second tenant (saves ~50% cost). Solution documented and shared.
  • TODO Onboarding call 3 with Basata — Scheduled.

Duty Support

  • TODO Support Sasha mybacs — New support request.

Personal / Tooling

  • DONE Smartclass v0.2.4 release — Fixed 6 bugs from code review: non-contiguous question ID validation, missing FK cascades, non-atomic exercise writes, test isolation (isolatedStorage: true), R2 stream consumption, and schema docs. (Release)
  • DONE Add shadcn/ui skills — Integrated shadcn/ui skill from the official repo into the agent-skills system. Synced remote skill files and created symlink. (PR #13)
  • DONE Add 1-on-1 automation report in Logseq — Set up the 1on1-report.sh automation for generating retrospective reports from Logseq notes.
  • IN-PROGRESS Complete Brazilian dashboard on Metabase for Yen — Dash 1 (Sales Overview) done; Dash 2 (Shipping Overview) and demo video still TODO.

Learning & Notes

  • Risk-based testing in dbt — Read Vinted Engineering article on tagging dbt models with high/low risk and frequency, then running tests selectively based on those tags.
  • Claude Skill Building Guide — Read Anthropic’s guide. Key takeaway: skill description should cover what it does, when to use, when not to use.
  • AI and labor — Reflected on the Oxygen Not Included DLC analogy: bionic dupes (robots) consume less oxygen and survive harsh environments but require lots of power. Like AI in our era — not ideal for early/simple work, but excellent for extending reach. When AI handles its strengths well, humans should double down on theirs.
  • 1-on-1 meetings — Read notes from The Manager’s Path: 1-on-1s are a chance to nurture the team/organization’s human resource.
  • Work ownership — “Làm việc bằng cái tâm” — treating company work as your own leads to better responsibility, empathy, and quality in everything.
  • Presales self-reflection — Calls are not yet at expected quality. Plan: replay past calls led by Huy/Vincent, pause at prospect questions, answer first, then watch how they handled it, and take learning points.

Next Week

  • P1 Onboarding call 3 with Basata — Presales, scheduled.
  • P1 Support Sasha mybacs — Duty Support, new request.
  • P2 Ingest Calendly data — Internal, prerequisite for phase 2 MRR analysis.
  • P2 Prepare for phase 2 MRR problem — Internal, depends on Calendly ingestion.
  • P2 Review and make a plan for MRR lead funnel / sales motion dimension — Internal, carry-over.
  • P2 Check back Census usage estimate (free tier confirmation) — Internal, noted for Mar 26.
  • P2 DAT-555 resolution — Internal, carry-over. Decide whether to remove fct_job_queue_performance and write Notion doc for team review.
  • P2 Fix excluding internal testing Zoho accounts — Internal, backlog item (DAT-524).
  • P2 Document Medenterprises performance use case — Duty Support, backlog (was due Feb 5).
  • P2 Brazilian dashboard Dash 2 + demo video for Yen — Teaching, carry-over.
  • P3 Search for a YouTube summarizer tool — Personal, backlog.

Career & Personal Consulting

  • High output on infrastructure and process work: Census migration, PR #842 fix, dbt test triage, PR Workflow comments, and 1-on-1 report automation — all completed this week. This is solid foundational work, but it’s predominantly maintenance/infra. The MRR analysis (the higher-leverage, revenue-impacting project) progressed only on understanding the data flow. Next week, carve out focused time for the Calendly ingestion and phase 2 planning.
  • Presales is growing: You handled the Showbie embedded dashboard QA use case end-to-end — from scoping requirements to discussing with the team and arriving at a cost-effective solution. That’s a full presales cycle for a technical use case. Keep building this muscle.
  • Delegation thinking is maturing: The AI-for-data-ops-bot idea and the explicit framework (pick task → resolve → document → let agent/junior try → iterate) shows good leadership instinct. Follow through by picking one DAT issue next week and running the first cycle with Thuan or an agent skill.
  • Self-awareness on call quality: Identifying that your presales calls aren’t at expected quality and devising a concrete practice plan (replay + pause + compare) is a strong move. Prioritize doing this for at least one past call before the Basata onboarding.
  • Context switches remain high: This week touched Internal, Presales, Duty Support, Personal, Teaching, and Smartclass. The 1-on-1 with anh Triet also surfaced concerns about pipeline complexity and PR Workflow adoption — both requiring sustained attention. Consider limiting to 2–3 project contexts per day next week.