Distributed Systems.
Telemetry.
Platform Infrastructure.
THOMPSON PLATFORM ENGINEERING LTD delivers senior backend engineering, telemetry pipelines, AWS serverless infrastructure, observability, and distributed systems consultancy for cloud-native environments.
12+
Years Engineering
10M+
Events Per Day
AWS
Serverless Systems
Python
Primary Stack
CAPABILITIES
Core Consultancy Areas
Backend & Platform Engineering
Python-first backend services, AWS serverless architectures, API design, and production-grade distributed systems.
Telemetry & Data Pipelines
High-throughput event ingestion, structured telemetry extraction, DynamoDB and S3 storage patterns at scale.
Observability & Reliability
Structured logging, CloudWatch monitoring, alerting across async microservice architectures, and incident response tooling.
Infrastructure & Delivery
Terraform, Docker, CI/CD pipelines, deployment automation, and operational reliability engineering.
SELECTED WORK
SignalForge — A Telemetry Intelligence Platform
A production-shaped analytics, detection and replay control plane for a retail device fleet, built end to end across five repositories. It is engineered around a single hard requirement — deterministic replay — so an archived event stream can be reprocessed through the same code to regenerate every derived output, byte for byte.
Event-time streaming core
Watermark-driven windowing with epoch-aligned boundaries and late-event repair, built function-shaped so the same pipeline runs in streaming, batch, and replay without modification.
Detection & alerting
Device-offline, store-outage and threshold-anomaly detectors feeding severity-aware alert routing, with replay-stable identity so a re-run never re-pages live on-call.
Datasets & dashboards
Partitioned S3/Parquet datasets with schema-per-file evolution, and DynamoDB-backed materialised projections for sub-5-second dashboard freshness.
Deterministic replay
A replay driver that reprocesses an archived stream through replay-isolated sinks, proven by an integration test asserting byte-identical output across the whole control plane.
7 / 7
Phases delivered to v1.0.0, each tagged and reviewed
384
Tests; mypy --strict and ruff clean throughout
50k / 8k
Devices and stores in the platform's design brief
2 yr
Retention horizon; sub-5s dashboard freshness target
TECHNOLOGY