Privacy engineering means getting your systems, not only your policies, to treat personal data correctly. We find where personal data actually sits, tag it with purpose and owner, map how it moves, then build consent sync, request handling and scheduled deletion into your own stack. Every step leaves records your team can show to customers, auditors or a regulator.
Key takeaways
- Most organisations hold personal data in far more places than their policies describe: logs, exports, test copies and SaaS tools.
- Discovery comes before automation, so the automation reaches every system that actually holds data.
- Privacy controls belong in the engineering lifecycle: design reviews, schema checks and test data rules.
- Each automated step should write evidence, so you can show what happened rather than describe it.
- We build inside your stack with your engineers, so the controls stay with your team.
Ask a team where its customers' phone numbers are stored and you will usually get one answer: the main database. Look closer and they are also in support tickets, application logs, a marketing export from last year, the analytics warehouse and a staging copy someone restored to debug a bug. Each copy is a cost, a security exposure and a question you cannot answer quickly when a customer, an enterprise buyer or a regulator asks.
Privacy engineering closes that gap. It treats personal data as something your systems should know about and act on, the same way they handle money or credentials.
Why privacy has become an engineering problem
Several pressures now land on the engineering team at once:
- Enterprise sales. Security and privacy questionnaires ask where data lives, who can reach it and how deletion works. Vague answers slow deals.
- Incidents. When something goes wrong, the first question is which records sat in the affected system. Without a map, the honest answer is everything.
- AI projects. Teams want to feed documents and records into AI tools, and nobody can say which of them hold personal data. Our data readiness for AI work starts from the same inventory.
- Regulation. India's DPDP Act is one driver among several, alongside sector regulators and the laws of other countries your customers live in.
- Cost. Data kept forever has to be stored, backed up and protected forever.
A policy cannot answer any of these on its own. A system that knows what it holds can.
How we run a privacy engineering engagement
We work in five steps, each feeding the next. Starting with a tool before you know what exists tends to produce automation that covers only the systems someone remembered.
Scan every system for personal data
Tag each field: what it is, why, whose
System to purpose to vendor
Consent, rights and deletion run themselves
Logs and reports you can hand over
- Discover. We combine system inventories, short interviews with owners and read-only scanning that runs inside your environment. Scans look at databases, object storage, log stores, SaaS exports and shared drives, and report metadata rather than copying records out.
- Classify. Each finding gets a small set of machine-readable tags: what kind of data it is, whose it is, why you hold it, who owns it and how sensitive it is. Tags live in your catalogue and storage labels, where code can read them.
- Map. We connect systems to purposes, vendors and retention rules, and tie the map to change management so a new table or a new vendor triggers an update.
- Automate. With the map in place, we build the repeatable parts: consent changes pushed to every downstream tool, request handling that fans out to the right systems, and retention jobs that delete or anonymise on schedule.
- Prove. Every automated step writes a minimal record: IDs, counts, times and outcomes, never the personal data itself. Reconciliation reports flag any system that has drifted from the map.
Privacy inside the development lifecycle
Discovery cleans up the past. These controls stop the mess coming back:
- Design review. A short checklist for new features: what personal data, for what purpose, kept how long, shared with whom.
- Schema and pull request checks. A new column that looks like an email address or phone number needs a tag before it merges.
- Logging rules. Log identifiers and tokens, not request bodies. Redact at the logging library, not by hand.
- Test data. Synthetic or masked data for non-production environments, and a check that blocks unmasked restores.
- Collect less. The cheapest data to protect is data you never stored. We look for fields nobody reads.
Where your data sits geographically is a related question, covered by our sovereign cloud and data residency service.
What you get
| Deliverable | What it is | Who uses it |
|---|---|---|
| Personal data inventory | Systems, stores and tools holding personal data, with confirmed scan findings | Engineering, security, privacy lead |
| Tagging scheme and applied tags | Data type, subject, purpose, owner and sensitivity, in your catalogue and storage | Data and platform teams |
| Data map | System to purpose to vendor to retention rule, linked to change management | Privacy lead, counsel, sales engineering |
| Automation | Consent sync, request workflow and retention jobs, built in your stack with tests | Engineering and operations |
| Lifecycle controls | Design checklist, pull request checks, logging rules, test data masking | Every product team |
| Evidence and runbooks | Job logs, reconciliation reports and how-to guides for your team | Auditors, enterprise customers, internal audit |
The goal is controls your team can run and change without us.
How to start
Pick three or four systems that hold the most customer data, plus a log platform and a shared folder, and we will run a pilot discovery with read-only access. The results usually show quickly how far the real picture is from the documented one, and that sets the scope for the rest. To set up a pilot, talk to us.
The DPDP side of this
Our DPDP practice, DPDP Ready, has a detailed guide on how this work maps to India's data protection law. Read it on DPDP Ready →
Frequently asked questions
What does a privacy engineer actually do?
A privacy engineer turns privacy requirements into working software and infrastructure. That means finding and classifying personal data, designing schemas and pipelines that collect less, building consent and deletion flows, and adding checks to the development process. The role sits between engineering, security and legal, and its output is code and configuration rather than documents.
What is the difference between privacy engineering and privacy compliance?
Compliance work decides what your organisation must do and writes it down: notices, policies, contracts. Privacy engineering makes systems do it, and produces evidence that they did. You need both, and they work best when the engineering team can read the policy as a list of testable behaviours.
How do you keep real customer data out of test and staging environments?
The usual pattern is to stop copying production data into lower environments by default, and to use synthetic data or masked extracts instead. Masking replaces names, phone numbers and identifiers with realistic fakes while keeping formats and relationships intact. A pipeline check can then block any restore of unmasked data into test.
Can privacy controls be added to an existing product without a rewrite?
In most cases, yes. Consent sync, request handling and retention jobs are usually built as services around your existing systems, connected through their APIs and event streams. The larger changes tend to be in logging and analytics, where personal data has often been captured without anyone deciding to.
Scope it with us
A 30-minute call to understand your systems and what good looks like. We aim to send a written, fixed-scope proposal within three working days.
Related services
This page describes engineering and security services. Where it mentions laws, rules or standards, it is general information, not legal advice.