Case studies

Software engineering case studies, measured in uptime, latency and load.

Selected work in healthcare, security, HRTech, retail and AI: the problem, what we changed and the measurable result.

Healthcare · Platform re-architecture

Patient scheduling: ~100 on-premise installs moved to a multi-tenant cloud.

A patient-management system with booking and SMS/email notifications ran as ~100 separate on-premise installs, costly to update and impossible to scale. We moved it to a multi-tenant AWS platform with a database per clinic, automated the migration, and later deployed it across multiple zones for expansion into new countries.

  • AWS
  • EC2
  • RDS
  • SQS
  • Multi-tenant
  • GDPR
Ask about this case →
99.99%
availability on critical workloads
Healthcare · AI document processing

Medical documents from scans, PDFs and free text, standardized with an LLM.

Lab results, discharge summaries and doctors' orders arrived as scans, PDFs and free text, in a different format and language from every laboratory. A mid-size LLM, chosen for speed and quality per cost, ran inside the client's AWS account and normalized them to LOINC at ~5,000 documents a month. A second, stronger model cross-checked every result, and any mismatch went to manual review and stayed out of the data until resolved. The structured data now powers patient analytics, test-history comparison and trend detection.

  • LLM
  • LOINC
  • AWS
  • LLM evaluation
  • Data normalization
Ask about this case →
>95%
extraction accuracy on our evaluation bench
Healthcare · Telemedicine platform

In-house WebRTC telemedicine for an offline clinic, integrated with its internal software.

An offline clinic needed online video consultations without depending on a third-party video provider for patient care. We built the WebRTC layer ourselves, with web and mobile clients, appointment booking and a patient portal integrated with the clinic's internal registration system. The doctor can switch a call to audio-only. Consultations are recorded for quality improvement, stored on AWS S3 and handled under the privacy policy and user consent. The clinic grew its consultation volume and can now bring in doctors remotely, without a visit to the clinic.

  • WebRTC
  • Web
  • Mobile
  • AWS S3
  • Privacy by design
Ask about this case →
Up to 30
concurrent video consultations
Healthcare · Systems integration

Two-way data exchange with laboratories and insurers that survives third-party outages.

The platform had to exchange test results and appointment records with laboratories and insurers, each with its own API and data model. We built integrations over FHIR and the providers' custom APIs, and made delivery reliable when a partner system goes down: SQS with a retry mechanism and the Transactional Outbox pattern, over secured connections. A mapping layer translates between provider formats, and integrations automated processes that were manual before.

  • FHIR
  • SQS
  • Transactional Outbox
  • Retries
  • Data mapping
Ask about this case →
Security · ID protection

White-label identity protection platform, built from zero and GDPR / ISO 27001 compliant.

We built a white-label product that monitors the dark web and public registries for exposed personal data (emails, passports, cards, driver licenses, ID and phone numbers) for both consumers and businesses. It covers monitoring, alerts, a user account, a partner API and an admin console, on a microservice architecture. Data is encrypted and never stored in plain text, with isolated access to it.

  • GDPR
  • ISO 27001
  • Microservices
  • White-label
  • Encryption
Ask about this case →
7M
personal data items monitored for ~1M users
Security · Zero-trust & certification

Live production platform moved to zero-trust with no downtime, and ISO 27001 certified.

To meet security compliance and pass ISO 27001 certification, we migrated an existing identity protection platform to a zero-trust model: verification on every request, least-privilege access, network segmentation, service isolation and protected administrator access. The move was gradual and without stopping the system. The audit raised findings, we fixed them, and the client passed the audit and certification.

  • Zero-trust
  • ISO 27001
  • Audit logging
  • Zero downtime
Ask about this case →
ISO 27001
audit and certification passed
Security · Identity & access

Sign-in and access control that holds up across products: SSO, OIDC, MFA and roles.

Built on many projects, for end users, client staff and partners, mostly from scratch and sometimes replacing an older scheme in a live system: SSO, OAuth / OIDC, MFA, role management, session handling and alternative sign-in methods on shared services, with access separated between partners and clients in multi-tenant products. The hard parts are consistent: migrating users without losing their access, edge cases in the role model, and session security.

  • SSO
  • OAuth / OIDC
  • MFA
  • RBAC
  • Multi-tenant
Ask about this case →
Security · Data protection

Sensitive data and secrets: encryption, erasure and a secure CI/CD pipeline.

On several projects with millions of records we encrypted, masked and anonymized sensitive data in production, implemented the right to erasure and retention policies, and kept sensitive data out of analytics, logs and copies. We also set up secure secret storage, injection at build and deploy time, and automated rotation of passwords, tokens and certificates. The hard parts: finding every place data lives, searching and filtering encrypted data, and finding secrets already leaked into logs and git history.

  • Encryption
  • Masking
  • Right to erasure
  • Secrets rotation
  • CI/CD
Ask about this case →
Security · Abuse protection

Keeping bots, credential guessing and API abuse out of public platforms.

On many projects we protected public sites, registration flows and partner APIs against bots, mass registrations, password guessing and API abuse, using rate limiting, bot detection and a mix of our own rules and managed edge protection from cloud and CDN vendors. The result was less fraud, lower load on the systems and no irrelevant bot data in them.

  • Rate limiting
  • Bot detection
  • WAF
  • Credential stuffing
Ask about this case →
HRTech · Platform engineering

The foundations of a multi-tenant HR platform: isolated tenants, reporting and payroll basics.

On a multi-tenant HR platform every company has its own database, provisioned automatically in about 10 seconds with schema updates handled by deployment, and analytics streamed into BigQuery through Pub/Sub. On top of it we built a Report Builder where a company picks almost any field from any module, saves reports and controls access, plus leave, time tracking and base-rate payroll. The hard parts were joining data from different modules into one result set and combining modules that all affect the final payout. The trade-off of the architecture is higher infrastructure cost and manual scaling.

  • Multi-tenant
  • Database per tenant
  • Report Builder
  • Time tracking
  • BigQuery
Ask about this case →
~10 s
to provision a new company's database and apply migrations
HRTech · Marketplace platform

A national job board built from scratch as a full SaaS product.

We built the whole job board platform from zero: vacancy publishing, search and filters, applications, resumes, job alerts, paid services for employers and moderation. Search runs on Elasticsearch and SEO was engineered separately for the large volume of vacancy pages. Response time was the hardest problem, solved with a combination of caching, database optimization and a read replica. Bot protection and moderation of companies and users keep spam and fake listings out. The platform earns through paid postings, subscriptions and ads.

  • Elasticsearch
  • SEO
  • Moderation
  • Bot protection
  • SaaS
Ask about this case →
100s
of vacancies and resumes posted per day
HRTech · Product build

A hiring module with AI, built from zero inside a multi-tenant HR platform.

We built the Hiring module of a multi-tenant HR platform from scratch: vacancies, candidate cards, the hiring funnel and statuses, evaluation, communication and reporting. On top of it we added resume parsing, an AI candidate analyzer and AI ranking, calendar integration, and automated emails through mail services. Anyone with the right permissions in any company on the platform can use it, so the client got hiring and the rest of HR in one all-in-one platform, with the routine work automated.

  • Multi-tenant
  • AI ranking
  • Resume parsing
  • Calendar integration
  • Email automation
Ask about this case →
Retail · Price analytics

Price comparison and analytics across the stores of a large retail chain.

For a large retail chain we built an application that compares and analyzes prices for the same products across its stores, using daily exports from the stores' cash register systems. Items are matched across stores by the name printed on the receipt, with an LLM check on top, and the application offers reports, dashboards, price history and alerts on discrepancies. It integrates with the company's internal systems. The hard parts were matching products and calculating the analytics. The results feed the company's internal accounting.

  • Price analytics
  • POS data
  • Product matching
  • LLM verification
  • Dashboards
Ask about this case →
~200
stores compared across ~3,000 products
AI · LLM cost & observability

Cutting LLM cost by 60% in production while keeping quality, with full observability.

We run LLM workloads through OpenRouter and track regular and reasoning tokens, model response time and whether answers match the expected schema. Cost came down through cheaper models where they are good enough, caching and prompt corrections, and quality is checked by a second model plus test benches that run the same data through different models many times. Queues and a fallback model keep the service up, and sensitive data is masked or replaced with placeholders before it reaches a model. The hard part was balancing price against quality.

  • OpenRouter
  • Token tracking
  • Model evaluation
  • Fallback models
  • Data masking
Ask about this case →
−60%
LLM cost with quality maintained
AI · Knowledge assistant (RAG)

An internal assistant that answers questions from 50,000 pages of technical manuals.

We built an internal knowledge assistant over about 50,000 pages of PDF installation, operation and service manuals for technical equipment, using vector search over a vector database and the client's own model. Every answer is verified by an LLM, and quality is measured against a test set of questions and answers. Search quality and document chunking were the hardest parts. The client got an assistant with a ready knowledge base of its equipment documentation.

  • RAG
  • Vector search
  • Vector database
  • LLM verification
  • Evaluation set
Ask about this case →
50,000
pages of PDF documentation searchable
AI · LLM process automation

LLM-powered automation of reports, email triage, documents and support on one shared infrastructure.

We built LLM-driven components that take over routine work: preparing reports, triaging incoming email, handling documents and answering support requests. They connect to databases, mail, a CRM and internal services, with different cloud models chosen for different tasks. They are deliberately constrained: they read data and issue standard answers based on specific templates, and they never change records. The hard part was building a single infrastructure for all of them, and the client got high-quality automation of its processes.

  • LLM
  • Process automation
  • Multi-model
  • Prompt engineering
  • CRM & email integration
Ask about this case →
Cross-domain · Messaging & notifications

A notification service that sends over 10,000 messages a day, with delays and retries.

On healthcare, HR and security products we built a service for mass email, SMS and push notifications: templates, personalization, scheduled delivery and recipient segmentation. It runs on queues with idempotent sending, retries and delayed delivery, tracks results through the Mailgun API and webhooks, and respects recipient consent and unsubscribes. The client got a service that sends a large volume of notifications on a delivery schedule, resends failed ones and reports analytics.

  • Email
  • SMS
  • Push
  • Queues
  • Idempotency
  • Mailgun
Ask about this case →
10k+
notifications per day

Facing a similar problem? Let's talk specifics.

Book a Technical Consultation →