Daniel F.-Senior Backend Engineer

A SaaS platform used by investment banks and private equity firms to run M&A transactions from diligence through close. The work covered the backend services on the company records underneath those deals: entity resolution against a large company database, record linkage from filings, model evaluation, and the validation and audit tooling that keeps automated decisions accurate and auditable.

• Owned the backend services behind entity resolution in the deal workflow, where a wrong match reaches bankers mid transaction. Took match recommendations into production write paths with full audit trails.
• Reviewed the product owner's initial spec, documented conflicts, recorded decisions in writing, then built and shipped the services.
• Labelled a sample dataset before a large migration. Found that only 60% was usable and 10% contained reversed labels. Fixed the data issues before service design.
• Added an LLM judge (Claude / Gemini) for cases the rules engine could not resolve. Measured 92.5% precision, found remaining errors from shared parent domains, and blocked auto approval until that pattern was fixed.
• Fixed bad data handling and added checks so the same errors cannot reappear. Each fix is recorded in the same transaction, with validation at the API and again at save.
• Mentored engineers to check that code matches the product and modelling rules, not only that it runs correctly.

Technologies: Python, FastAPI, C#, .NET, TypeScript, Dagster, PostgreSQL, Google Cloud (Cloud Run, Cloud SQL, Cloud Storage), Anthropic Claude API, Google Gemini API

An enterprise AI SaaS company with client-facing products across travel, automotive, and analytics. The work covered high-volume ingestion services and service infrastructure, cost-aware compute, and the CI/CD and evaluation standards used across the team.

• Ran the batch scoring and feature services behind client-facing dashboards, the numbers account managers quoted to clients. Workers for normalisation, currency conversion, and deduplication, scheduled on Airflow.
• Treated compute cost as a margin question, not just latency. Batching, profiling, and worker sizing. Runtimes down over 50%.
• Set the CI/CD and testing standard for the team so changes stopped reaching clients unverified. Pytest, ephemeral test environments, Git-based deployment.

Technologies: Python, aiothp, Apache Kafka, Apache Airflow, Elasticsearch, AWS S3, Docker, GitHub Actions

An enterprise conversational AI platform automating customer service across chat and voice.

• Built the services behind the conversation analytics clients used to judge whether the platform was earning its contract: ingestion endpoints, transformation workers, and the reporting API.
• Kept those endpoints responsive as client data grew. Query and index tuning, connection pooling, and schema design on PostgreSQL and MySQL.
• Stopped bad data at the service boundary before it reached reporting. Request validation, schema checks, and idempotent writes on every run.

Technologies: Python, FastAPI, PostgreSQL, MySQL, Docker, GitLab CI