Reesync AI Ltd
AWS Data Platform Rescue Audit
A fixed-price, two-week audit for UK teams running Glue, Redshift, S3, Athena and Tableau. Find out where the money is going, why the pipelines keep breaking, and what it will take to trust the reporting again.
UK IT Industry Award winner 2021 · 13+ years in regulated UK environments · Ran a ~500-user Tableau estate on AWS · Fixed price: £2,500 + VAT
The problem
Most AWS data platforms were not designed. They accumulated. A Glue job here, a Redshift cluster there, a few Lambdas someone wrote in 2021, and a Tableau estate sitting on top of all of it.
It works, mostly. But the bill keeps creeping up, the overnight jobs fail more often than anyone admits, and when a dashboard number looks wrong, nobody is completely sure whether it is the data, the pipeline or the definition.
The people who built it have often moved on. The people who run it are too busy keeping it alive to step back and fix it properly. That is the gap this audit fills.
Sound familiar?
- The monthly AWS bill keeps rising and nobody can explain exactly why
- Overnight Glue or Lambda jobs fail and get fixed by hand, again
- Dashboards take 30 seconds to load, so people export to Excel instead
- Finance found out the data was stale before your team did
- Two dashboards show two different numbers for the same KPI
- One person understands how the whole platform fits together
- Athena costs are “ad hoc” in every sense
- Nobody is sure what is safe to switch off
If three or more of these sound familiar, a conversation is probably worth 30 minutes of your time.
Download the AWS Platform Health Checklist (PDF)This is for you if:
- You run an AWS-based data or analytics platform using at least two of Glue, Redshift, S3, Athena, Lambda or Step Functions
- You have Tableau, Power BI or QuickSight sitting on top of it
- You are a Head of Data, BI Manager, CTO, Engineering Manager or platform lead who has to answer for cost, reliability or reporting quality
- You want findings and a plan you can act on, not a slide deck
This is not for you if:
- Your estate is primarily Azure or GCP
- You want someone to guarantee savings before looking at any evidence
- You are looking for a free Well-Architected Review (AWS partners offer those, and they serve a different purpose)
- You need a 24/7 managed service (that is not what this is)
What the audit covers
Eight dimensions, one report
Cost optimisation
Where the spend actually goes: Glue DPUs, Redshift sizing, S3 storage classes, Athena scan volumes, idle and orphaned resources.
Reliability
Pipeline failure patterns, retry behaviour, manual reruns, single points of failure.
Operational excellence
Scheduling, environment separation, release and change control, runbooks.
Security and governance signals
IAM hygiene, long-lived keys, encryption indicators, public access flags, audit logging presence. Signals, not a penetration test.
Observability
Whether you find out about failures from alarms or from users.
Data engineering hygiene
Partitioning, job design, data quality checks, duplication, dead code and dead datasets.
BI and Tableau readiness
Extract strategy, refresh schedules, slow dashboards, row-level security, server health, KPI definition clarity.
Maintainability
Documentation, key-person risk, and what happens when the person who built it leaves.
How it works
Two weeks, four steps
Scope call (day 0)
We agree what is in scope, who I should talk to, and which access route suits your security posture. You choose how much access I get, including a no-access option.
Evidence collection (days 1–3)
Metadata only. Configurations, job histories, cost data, schedules, logs and screenshots. Never table contents, never customer records. Plus 30–45 minute interviews with two or three of your people.
Analysis and scoring (days 4–8)
Every finding gets evidence, a severity, a business impact, an effort estimate and a confidence score. No vague observations.
Report and playback (days 9–10, plus playback)
You get a board-ready executive report, a scored findings register and a 30/60/90-day remediation plan, walked through in a 90-minute playback session. Then 30 days of email questions included.
Your team's total time commitment: about 3–4 hours.
The point is to take work off your plate, not add to it.
Sample findings
The kind of thing the audit finds
Illustrative finding 1: avoidable compute cost
Hourly Glue jobs scanning unpartitioned S3 data, re-processing everything on every run. Around £1,400/month in avoidable DPU and Athena scan cost, fixable in days.
Illustrative finding 2: silent failures
Overnight pipeline with no failure alerting. Finance discovers stale data at 9am; the team finds out from finance. One CloudWatch alarm and an alert route would have caught every instance.
Illustrative finding 3: BI load nobody asked for
Tableau extracts refreshing four times a day against Redshift during business hours, driving both slow dashboards and query queuing for everyone else. A schedule and extract redesign fixes both at once.
These are illustrative examples based on patterns seen across real AWS estates, not findings from a named client.
Sample report
See exactly what you get
Download the full sample report for Acme Utilities Ltd, a synthetic client with a realistically messy AWS estate. Every section, every score, every recommendation, exactly as a real report would look.
Download the sample report (PDF)Packages and pricing
Start small or go straight to the audit
Platform Snapshot
£495 + VATA one-hour working session plus a short written memo covering your top five issues, based on a questionnaire and your cost data. No account access needed. Delivered within five working days. The full £495 is credited against an audit booked within 30 days.
Rescue Audit
£2,500 + VATThe full two-week audit across all eight dimensions. Executive report, scored findings register, 30/60/90-day plan, playback session and 30 days of follow-up questions. Fixed price, fixed scope, no surprises.
Remediation Sprint
from £7,500 + VATA two-to-four week fixed-scope sprint that fixes the highest-value findings from the audit. Before/after evidence, updated runbooks and a proper handover. Scoped from your audit, so you only pay to fix what is proven.
Ongoing support
from £1,500/month + VATFor teams that want a monthly health check, advisory time and a steady stream of small improvements after the audit. Clear boundaries, not a 24/7 managed service.
Fair warning on the audit price: if within the first two days I conclude the audit will not surface material value for you, I will say so and you pay nothing.
Why me
Built these platforms. Ran them. Fixed them.
I spent 13+ years at a regulated UK utility doing exactly this work: building AWS data pipelines with Glue, PySpark and Redshift, running the enterprise Tableau Server estate for around 500 users, and leading the Data Engineering Centre of Excellence.
The Fatigue Manager project I led won Automation Project of the Year at the UK IT Industry Awards in 2021. The Oracle to Redshift migration work I delivered sat at the core of a programme estimated to save around £3m a year. The Fleet MI platform I built removed around £60k a year of legacy reporting cost.
So when I look at your Glue jobs, your Redshift cluster or your Tableau Server, I am not pattern-matching from a checklist. I have run estates like yours in environments where audit trails, governance and data accuracy were non-negotiable.
Security and access
What I access, and what I never touch
The audit works on metadata: configurations, schedules, job run histories, cost data, log patterns and screenshots. It does not need your business data.
You choose one of three access routes:
- —No access. You complete a questionnaire and provide exports and screenshots. I never touch your account.
- —You run the script. A read-only collection script you run yourself, producing a redacted metadata bundle you review before sending.
- —Read-only role. A cross-account read-only IAM role deployed via CloudFormation, locked down with an external ID, scoped to least privilege, and revocable by you at any time.
What I never access
Table contents, customer records, credentials or secrets. Nothing is changed in your environment during an audit. Client material is processed locally by default, and AI tooling is only applied to your material with your written agreement. An NDA is available before any conversation about your estate.
Reesync AI Ltd is a UK registered company, VAT registered, with professional indemnity insurance.
FAQ
Common questions
Two weeks from now, you could know exactly what to fix
Book a 20-minute triage call. If the audit is not the right fit, I will tell you on the call and point you somewhere more useful.
Or email hello@reesyncai.com with a sentence about your estate, and I will reply with an honest view.
Get in touch
What is your AWS data platform doing that it should not be?
Send a short message about your platform. I'll tell you whether the audit is a good fit or whether something else would serve you better.
What is your AWS data platform doing that it should not be?